Abstract Time series forecasting plays a critical role in numerous real-world applications, such as finance, healthcare, transportation, and scientific computing. In recent years, deep learning has become a powerful tool for modeling complex temporal patterns and improving forecasting accuracy. This survey provides an overview of recent deep learning approaches for time series forecasting, involving various architectures including RNNs, CNNs, GNNs, transformers, large language models, MLP-based models, and diffusion models. We first identify key challenges in the field, such as temporal dependency, efficiency, and cross-variable dependency, which drive the development of forecasting techniques. Then, the general advantages and limitations of each architecture are discussed to contextualize their adaptation in time series forecasting. Furthermore, we highlight promising design trends like multi-scale modeling, decomposition, and frequency-domain techniques, which are shaping the future of the field. This paper serves as a compact reference for researchers and practitioners seeking to understand the current landscape and future trajectory of deep learning in time series forecasting.

Time series anomaly detection is a critical machine learning task for numerous applications, such as finance, healthcare, and industrial systems. However, even high-performing models may exhibit potential issues such as biases, leading to unreliable outcomes and misplaced confidence. While model explanation techniques, particularly visual explanations, offer valuable insights by elucidating model attributions of their decision, many limitations still exist—They are primarily instance-based and not scalable across the dataset, and they provide one-directional information from the model to the human side, lacking a mechanism for users to address detected issues. To fulfill these gaps, we introduce HILAD , a novel framework designed to foster a dynamic and bidirectional collaboration between humans and AI for enhancing anomaly detection models in time series. Through our visual interface, HILAD empowers domain experts to detect, interpret, and correct unexpected model behaviors at scale. Our evaluation through user studies with two models and three time series datasets demonstrates the effectiveness of HILAD , which fosters a deeper model understanding, immediate corrective actions, and model reliability enhancement.

Real-time path tracing is rapidly becoming the standard for rendering in entertainment and professional applications. In scientific visualization, volume rendering plays a crucial role in helping researchers analyze and interpret complex 3D data. Recently, photorealistic rendering techniques have gained popularity in scientific visualization, yet they face significant challenges. One of the most prominent issues is slow rendering performance and high pixel variance caused by Monte Carlo integration. In this work, we introduce a novel radiance caching approach for path-traced volume rendering. Our method leverages advances in volumetric scene representation and adapts 3D Gaussian splatting to function as a multi-level, path-space radiance cache. This cache is designed to be trainable on the fly, dynamically adapting to changes in scene parameters such as lighting configurations and transfer functions. By incorporating our cache, we achieve less noisy, higher-quality images without increasing rendering costs. To evaluate our approach, we compare it against a baseline path tracer that supports uniform sampling and next-event estimation and the state-of-the-art for neural radiance caching. Through both quantitative and qualitative analyses, we demonstrate that our path-space radiance cache is a robust solution that is easy to integrate and significantly enhances the rendering quality of volumetric visualization applications while maintaining comparable computational efficiency.

Understanding and distinguishing temporal patterns in time series data is essential for scientific discovery and decision-making. For example, in biomedical research, uncovering meaningful patterns in physiological signals can improve diagnosis, risk assessment, and patient outcomes. However, existing methods for time series pattern discovery face major challenges, including high computational complexity, limited interpretability, and difficulty in capturing meaningful temporal structures. To address these gaps, we introduce a novel learning framework that jointly trains two Transformer models using complementary time series representations: shapelet-based representations to capture localized temporal structures and traditional feature engineering to encode statistical properties. The learned shapelets serve as interpretable signatures that differentiate time series across classification labels. Additionally, we develop a visual analytics system-SigTime-with coordinated views to facilitate exploration of time series signatures from multiple perspectives, aiding in useful insights generation. We quantitatively evaluate our learning framework on eight publicly available datasets and one proprietary clinical dataset. Additionally, we demonstrate the effectiveness of our system through two usage scenarios along with the domain experts: one involving public ECG data and the other focused on preterm labor analysis.

Ensemble datasets are ever more prevalent in various scientific domains. In climate science, ensemble datasets are used to capture variability in projections under plausible future conditions including greenhouse and aerosol emissions. Each ensemble model run produces projections that are fundamentally similar yet meaningfully distinct. Understanding this variability among ensemble model runs and analyzing its magnitude and patterns is a vital task for climate scientists. In this paper, we present ClimateSOM, a visual analysis workflow that leverages a self-organizing map (SOM) and Large Language Models (LLMs) to support interactive exploration and interpretation of climate ensemble datasets. The workflow abstracts climate ensemble model runs-spatiotemporal time series-into a distribution over a 2D space that captures the variability among the ensemble model runs using a SOM. LLMs are integrated to assist in sensemaking of this SOM-defined 2D space, the basis for the visual analysis tasks. In all, ClimateSOM enables users to explore the variability among ensemble model runs, identify patterns, compare and cluster the ensemble model runs. To demonstrate the utility of ClimateSOM, we apply the workflow to an ensemble dataset of precipitation projections over California and the Northwestern United States. Furthermore, we conduct a short evaluation of our LLM integration, and conduct an expert review of the visual workflow and the insights from the case studies with six domain experts to evaluate our approach and its utility.

Financial literacy empowers individuals to make informed and effective financial decisions, improving their overall financial well-being and security. However, for many people understanding financial concepts can be daunting and only half of US adults are considered financially literate. Data visualization simplifies these concepts, making them accessible and engaging for learners of all ages. This systematic review analyzes 37 research papers exploring the use of data visualization and visual analytics in financial education and literacy enhancement. We classify these studies into five key areas: (1) the evolution of visualization use across time and space, (2) motivations for using visualization tools, (3) the financial topics addressed and instructional approaches used, (4) the types of tools and technologies applied, and (5) how the effectiveness of teaching interventions was evaluated. Furthermore, we identify research gaps and highlight opportunities for advancing financial literacy. Our findings offer practical insights for educators and professionals to effectively utilize or design visual tools for financial literacy. • This systematic review provides the first structured synthesis of data visualization in financial literacy education, analyzing 37 studies by addressing five key research questions on its historical development, research motivations, financial education pedagogy, visualization technologies, and intervention evaluation. • We identify critical research gaps, including the need for more interactive and adaptive visualization tools, broader audience inclusion beyond students, and stronger empirical validation of visualization effectiveness. These findings offer practical guidance for educators, researchers, and tool developers to enhance financial literacy education.

The tsNET algorithm adapts the popular dimensional reduction method t-SNE for graph drawing to compute high-quality drawings, preserving the neighborhood and clustering structure. However, its O(nm) runtime results in poor scalability for large graphs. In this poster, we present three fast algorithms for reducing the time complexity of tsNET to O(n log n) time and O(n) time, by integrating new fast methods for computation of high-dimensional probabilities and entropy computation with fast t-SNE algorithms for computation of KL divergence gradient. Specifically, we present two O(n log n)-time algorithms BH-tsNET and FIt-tsNET, incorporating partial BFS-based high-dimensional probability computation and a new quadtree-based entropy computation with fast t-SNE algorithms, and O(n)-time algorithm L-tsNET, introducing a new fast interpolation-based entropy computation. Extensive experiments using benchmark data sets confirm that BH-tsNET, FIt-tsNET, and L-tsNET outperform tsNET, achieving 93.5%, 96%, and 98.6% faster runtime, respectively, while computing similar quality drawings in terms of quality metrics (neighborhood preservation, stress, shape-based metrics, and edge crossing) and visual comparison.

Training-free open-vocabulary semantic segmentation (OVS) aims to segment images given a set of arbitrary textual categories without costly model fine-tuning. Existing solutions often explore attention mechanisms of pre-trained models, such as CLIP, or generate synthetic data and design complex retrieval processes to perform OVS. However, their performance is limited by the capability of reliant models or the suboptimal quality of reference sets. In this work, we investigate the largely overlooked data quality problem for this challenging dense scene understanding task, and identify that a high-quality reference set can significantly benefit training-free OVS. With this observation, we introduce a data-quality-oriented framework, comprising a data pipeline to construct a reference set with well-paired segment-text embeddings and a simple similarity-based retrieval to unveil the essential effect of data. Remarkably, extensive evaluations on ten benchmark datasets demonstrate that our method outperforms all existing training-free OVS approaches, highlighting the importance of data-centric design for advancing OVS without training. Our code is available at https://github.com/xiweix/ReME .

Environmental experts have developed the DPSIR (Driver, Pressure, State, Impact, Response) framework to systematically study and communicate key relationships between society and the environment. Using this framework requires experts to construct a DPSIR taxonomy from a corpus, annotate the documents, and identify DPSIR variables and relationships, which is laborious and inflexible. Automating it with conventional text mining faces technical challenges, primarily because the taxonomy often begins with abstract definitions, which experts progressively refine and contextualize as they annotate the corpus. In response, we develop GreenMine, a system that supports interactive text mining with prompt engineering. The system implements a prompting pipeline consisting of three simple and evaluable subtasks. In each subtask, the DPSIR taxonomy can be defined in natural language and iteratively refined as experts analyze the corpus. To support users evaluate the taxonomy, we introduce an uncertainty score based on response consistency. Then, we design a radial uncertainty chart that visualizes uncertainties and corpus topics, which supports interleaved evaluation and exploration. Using the system, experts can progressively construct the DPSIR taxonomy and annotate the corpus with LLMs. Using real-world interview transcripts, we present a case study to demonstrate the capability of the system in supporting interactive mining of DPSIR relationships, and an expert review in the form of collaborative discussion to understand the potential and limitations of the system. We discuss the lessons learned from developing the system and future opportunities for supporting interactive text mining in knowledge-intensive tasks for other application scenarios.

Abstract We present HyperFLINT (Hypernetwork‐based FLow estimation and temporal INTerpolation), a novel deep learning‐based approach for estimating flow fields, temporally interpolating scalar fields, and facilitating parameter space exploration in spatio‐temporal scientific ensemble data. This work addresses the critical need to explicitly incorporate ensemble parameters into the learning process, as traditional methods often neglect these, limiting their ability to adapt to diverse simulation settings and provide meaningful insights into the data dynamics. HyperFLINT introduces a hypernetwork to account for simulation parameters, enabling it to generate accurate interpolations and flow fields for each timestep by dynamically adapting to varying conditions, thereby outperforming existing parameter‐agnostic approaches. The architecture features modular neural blocks with convolutional and deconvolutional layers, supported by a hypernetwork that generates weights for the main network, allowing the model to better capture intricate simulation dynamics. A series of experiments demonstrates HyperFLINT's significantly improved performance in flow field estimation and temporal interpolation, as well as its potential in enabling parameter space exploration, offering valuable insights into complex scientific ensembles.

Egocentric networks, often visualized as node-link diagrams, portray the complex relationship (link) dynamics between an entity (node) and others. However, common analytics tasks are multifaceted, encompassing interactions among four key aspects: strength, function, structure, and content. Current node-link visualization designs may fall short, focusing narrowly on certain aspects and neglecting the holistic, dynamic nature of egocentric networks. To bridge this gap, we introduce SpreadLine, a novel visualization framework designed to enable the visual exploration of egocentric networks from these four aspects at the microscopic level. Leveraging the intuitive appeal of storyline visualizations, SpreadLine adopts a storyline-based design to represent entities and their evolving relationships. We further encode essential topological information in the layout and condense the contextual information in a metro map metaphor, allowing for a more engaging and effective way to explore temporal and attribute-based information. To guide our work, with a thorough review of pertinent literature, we have distilled a task taxonomy that addresses the analytical needs specific to egocentric network exploration. Acknowledging the diverse analytical requirements of users, SpreadLine offers customizable encodings to enable users to tailor the framework for their tasks. We demonstrate the efficacy and general applicability of SpreadLine through three diverse real-world case studies (disease surveillance, social media trends, and academic career evolution) and a usability study.

Recent advancements in Large Language Models (LLMs) and Prompt Engineering have made chatbot customization more accessible, significantly reducing barriers to tasks that previously required programming skills. However, prompt evaluation, especially at the dataset scale, remains complex due to the need to assess prompts across thousands of test instances within a dataset. Our study, based on a comprehensive literature review and pilot study, summarized five critical challenges in prompt evaluation. In response, we introduce a feature-oriented workflow for systematic prompt evaluation. In the context of text summarization, our workflow advocates evaluation with summary characteristics (feature metrics) such as complexity, formality, or naturalness, instead of using traditional quality metrics like ROUGE. This design choice enables a more user-friendly evaluation of prompts, as it guides users in sorting through the ambiguity inherent in natural language. To support this workflow, we introduce Awesum, a visual analytics system that facilitates identifying optimal prompt refinements for text summarization through interactive visualizations, featuring a novel Prompt Comparator design that employs a BubbleSet-inspired design enhanced by dimensionality reduction techniques. We evaluate the effectiveness and general applicability of the system with practitioners from various domains and found that (1) our design helps overcome the learning curve for non-technical people to conduct a systematic evaluation of summarization prompts, and (2) our feature-oriented workflow has the potential to generalize to other NLG and image-generation tasks. For future works, we advocate moving towards feature-oriented evaluation of LLM prompts and discuss unsolved challenges in terms of human-agent interaction.

Sensemaking on a large collection of documents (corpus) is a challenging task often found in fields such as market research, legal studies, intelligence analysis, political science, or computational linguistics. Previous works approach this problem from topic- and entity-based perspectives, but the capability of the underlying NLP model limits their effectiveness. Recent advances in prompting with LLMs present opportunities to enhance such approaches with higher accuracy and customizability. However, poorly designed prompts and visualizations could mislead users into falsely interpreting the visualizations and hinder the system's trustworthiness. In this paper, we address this issue by taking into account the user analysis tasks and visualization goals in the prompt-based data extraction stage, thereby extending the concept of Model Alignment. We present HINTs, a VA system for supporting sensemaking on large collections of documents, combining previous entity-based and topic-based approaches. The visualization pipeline of HINTs consists of three stages. First, entities and topics are extracted from the corpus with prompts. Then, the result is modeled as a hypergraph and hierarchically clustered. Finally, an enhanced space-filling curve layout is applied to visualize the hypergraph for interactive exploration. The system further integrates an LLM-based intelligent chatbot agent in the interface to facilitate the sensemaking of interested documents. To demonstrate the generalizability and effectiveness of the HINTs system, we present two case studies on different domains and a comparative user study. We report our insights on the behavior patterns and challenges when intelligent agents are used to facilitate sensemaking. We find that while intelligent agents can address many challenges in sensemaking, the visual hints that visualizations provide are still necessary. We discuss limitations and future work for combining interactive visualization and LLMs more profoundly to better support corpus analysis.

Multivariate networks are commonly found in real-world data-driven applications. Uncovering and understanding the relations of interest in multivariate networks is not a trivial task. This paper presents a visual analytics workflow for studying multivariate networks to extract associations between different structural and semantic characteristics of the networks (e.g., what are the combinations of attributes largely relating to the density of a social network?). The workflow consists of a neural-network-based learning phase to classify the data based on the chosen input and output attributes, a dimensionality reduction and optimization phase to produce a simplified set of results for examination, and finally an interpreting phase conducted by the user through an interactive visualization interface. A key part of our design is a composite variable construction step that remodels nonlinear features obtained by neural networks into linear features that are intuitive to interpret. We demonstrate the capabilities of this workflow with multiple case studies on networks derived from social media usage and also evaluate the workflow with qualitative feedback from experts.

Implicit neural representations (INRs) have emerged as a powerful tool for compressing large-scale volume data. This opens up new possibilities for in situ visualization. However, the efficient application of INRs to distributed data remains an underexplored area. In this work, we develop a distributed volumetric neural representation and optimize it for in situ visualization. Our technique eliminates data exchanges between processes, achieving state-of-the-art compression speed, quality and ratios. Our technique also enables the implementation of an efficient strategy for caching large-scale simulation data in high temporal frequencies, further facilitating the use of reactive in situ visualization in a wider range of scientific problems. We integrate this system with the Ascent infrastructure and evaluate its performance and usability using real-world simulations.

Data slice finding is an emerging technique for validating machine learning (ML) models by identifying and analyzing subgroups in a dataset that exhibit poor performance, often characterized by distinct feature sets or descriptive metadata. However, in the context of validating vision models involving unstructured image data, this approach faces significant challenges, including the laborious and costly requirement for additional metadata and the complex task of interpreting the root causes of underperformance. To address these challenges, we introduce AttributionScanner, an innovative human-in-the-loop Visual Analytics (VA) system, designed for metadata-free data slice finding. Our system identifies interpretable data slices that involve common model behaviors and visualizes these patterns through an Attribution Mosaic design. Our interactive interface provides straightforward guidance for users to detect, interpret, and annotate predominant model issues, such as spurious correlations (model biases) and mislabeled data, with minimal effort. Additionally, it employs a cutting-edge model regularization technique to mitigate the detected issues and enhance the model's performance. The efficacy of AttributionScanner is demonstrated through use cases involving two benchmark datasets, with qualitative and quantitative evaluations showcasing its substantial effectiveness in vision model validation, ultimately leading to more reliable and accurate models.

The advances in multi-modal foundation models (FMs) (e.g., CLIP and LLaVA) have facilitated the auto-labeling of large-scale datasets, enhancing model performance in challenging downstream tasks such as open-vocabulary object detection and segmentation. However, the quality of FM-generated labels is less studied as existing approaches focus more on data quantity over quality. This is because validating large volumes of data without ground truth presents a considerable challenge in practice. Existing methods typically rely on limited metrics to identify problematic data, lacking a comprehensive perspective, or apply human validation to only a small data fraction, failing to address the full spectrum of potential issues. To overcome these challenges, we introduce VISTA, a visual analytics framework that improves data quality to enhance the performance of multi-modal models. Targeting the complex and demanding domain of open-vocabulary image segmentation, VISTA integrates multi-phased data validation strategies with human expertise, enabling humans to identify, understand, and correct hidden issues within FM-generated labels. Through detailed use cases on two benchmark datasets and expert reviews, we demonstrate VISTA's effectiveness from both quantitative and qualitative perspectives.

The ability to monitor and interpret hardware system events and behaviors is crucial to improving the robustness and reliability of these systems, especially in a supercomputing facility. The growing complexity and scale of these systems demand an increase in monitoring data collected at multiple fidelity levels and varying temporal resolutions. In this work, we aim to build a holistic analytical system that helps make sense of such massive data, mainly the hardware logs, job logs, and environment logs collected from disparate subsystems and components of a supercomputer system. This end-to-end log analysis system, coupled with visual analytics support, allows users to glean and promptly extract supercomputer usage and error patterns at varying temporal and spatial resolutions. We use multi-resolution dynamic mode decomposition (mrDMD), a technique that depicts high-dimensional data as correlated spatial-temporal variations patterns or modes, to extract variation patterns isolated at specified frequencies. Our improvements to the mrDMD algorithm help promptly reveal useful information in the massive environment log dataset, which is then associated with the processed hardware and job log datasets using our visual analytics system. Furthermore, our system can identify the usage and error patterns filtered at user, project, and subcomponent levels. We exemplify the effectiveness of our approach with two use scenarios with the Cray XC40 supercomputer.

Visualization plays a vital role in making sense of complex network data. Recent studies have shown the potential of using extended reality (XR) for the immersive exploration of networks. The additional depth cues offered by XR help users perform better in certain tasks when compared to using traditional desktop setups. However, prior works on immersive network visualization rely mostly on singular, static graph layouts to present the data to the user. This poses a problem since there is no optimal layout for all possible tasks. The choice of layout heavily depends on the type of network and the task at hand. We introduce a multi-layout design that promotes more efficient use of the available space in VR environments and allows users to explore hierarchical network data in immersive space effectively. We implement our design with a choice of four distinct views on the network. The resulting system leverages various existing layout techniques to efficiently use the available space in VR and provide an optimal view of the data depending on the task and the level of detail required to solve it. To evaluate our approach, we conducted a user study comparing it against the state of the art for immersive network visualization. Participants performed tasks at varying spatial scopes. The results show that our approach outperforms the baseline in spatially focused scenarios as well as when the whole network needs to be considered.

As machine learning (ML) gains wider adoption in real-world applications, the validation of ML models becomes fundamental for its productization, particularly in safety-critical applications. Recently, data slice finding has emerged as a popular method for validating ML models, but it requires additional metadata or cross-modal embeddings for the slices to be interpretable. We propose ConceptSlicer, an integrated workflow that facilitates the slicing of computer vision models using visual concepts. This approach breaks down the image dataset into interpretable visual concepts, serving as metadata in the slice finding process. Our system offers insights into model issues and enables a deeper understanding of computer vision models’ strengths and weaknesses. We evaluate ConceptSlicer through interviews with eight domain experts and machine learning practitioners, and fine-tune the ML models based on their feedback. Our study also highlights varied attitudes towards large foundational models, encouraging contemplation of the challenges and opportunities presented by this technological advancement.

Whereas rapidly increasing heterogeneous compute capabilities continue to facilitate further scalability, modern applications often instead get limited by suboptimal data movement, as more and more data needs to be shipped across different hardware components (i.e., CPUs, GPUs, and other types of accelerators). We posit that understanding and improving data movement in modern use-cases require a holistic understanding of the underlying Hardware usage as well as the Communication patterns within the overall context of the Application, or as we call it, the HAC domain. Collecting and correlating HAC data currently requires interacting with several profiling tools and libraries, resulting in a tedious workflow that is neither scalable nor portable. Furthermore, existing tools for visualizing data movement profiles also focus on these domains individually, rather than offering a holistic view.We present a unified framework for tracking and visualizing data movement trends in large-scale applications deployed on heterogeneous architectures. Our framework has two interoperable components. (1) DMTracker is a lean software layer that provides a simple interface for configurable HAC profiling of GPU-enabled applications and abstracts away the complexity in using several profiling tools, resulting in a streamlined and time-correlated event history across the HAC domains. (2) DMVis is a web-based tool that combines several linked visualizations to offer holistic visual insights into the runtime behavior and resources utilization of applications, proving pivotal in identifying computationally expensive tasks and data transfers across devices. In this paper, we present the design and prototype of our framework, developed in consultation with domain experts and demonstrated on two case studies, including one for a large language model training. Initial impressions from the experts indicate a positive turn in their usual workflow of observing and tuning the performance through improved data movement strategies.

Abstract Adaptive Mesh Refinement (AMR) is becoming a prevalent data representation for HPC, and thus also for scientific visualization. AMR data is usually cell centric (which imposes numerous challenges), complex, and generally hard to render. Recent work on GPU‐accelerated AMR rendering has made much progress towards real‐time volume and isosurface rendering of such data, but so far this work has focused exclusively on ray marching, with simple lighting models and without scattering events or global illumination. True high‐quality rendering requires a modified approach that is able to trace arbitrary incoherent paths; but this may not be a perfect fit for the types of data structures recently developed for ray marching. In this paper, we describe a novel approach to high‐quality path tracing of complex AMR data, with a specific focus on analyzing and comparing different data structures and algorithms to achieve this goal.

Volume data is commonly found in many scientific disciplines, like medicine, physics, and biology. Experts rely on robust scientific visualization techniques to extract valuable insights from the data. Recent years have shown path tracing to be the preferred approach for volumetric rendering, given its high levels of realism. However, real-time volumetric path tracing often suffers from stochastic noise and long convergence times, limiting interactive exploration. In this paper, we present a novel method to enable real-time global illumination for volume data visualization. We develop Photon Field Networks-a phase-function-aware, multi-light neural representation of indirect volumetric global illumination. The fields are trained on multi-phase photon caches that we compute a priori. Training can be done within seconds, after which the fields can be used in various rendering tasks. To showcase their potential, we develop a custom neural path tracer, with which our photon fields achieve interactive framerates even on large datasets. We conduct in-depth evaluations of the method's performance, including visual quality, stochastic noise, inference and rendering speeds, and accuracy regarding illumination and phase function awareness. Results are compared to ray marching, path tracing and photon mapping. Our findings show that Photon Field Networks can faithfully represent indirect global illumination within the boundaries of the trained phase spectrum while exhibiting less stochastic noise and rendering at a significantly faster rate than traditional methods.

When telling a data story, an author has an intention they seek to convey to an audience. This intention can be of many forms such as to persuade, to educate, to inform, or even to entertain. In addition to expressing their intention, the story plot must balance being consumable and enjoyable while preserving scientific integrity. In data stories, numerous methods have been identified for constructing and presenting a plot. However, there is an opportunity to expand how we think and create the visual elements that present the story. Stories are brought to life by characters; often they are what make a story captivating, enjoyable, memorable, and facilitate following the plot until the end. Through the analysis of 160 existing data stories, we systematically investigate and identify distinguishable features of characters in data stories, and we illustrate how they feed into the broader concept of "character-oriented design". We identify the roles and visual representations data characters assume as well as the types of relationships these roles have with one another. We identify characteristics of antagonists as well as define conflict in data stories. We find the need for an identifiable central character that the audience latches on to in order to follow the narrative and identify their visual representations. We then illustrate "character-oriented design" by showing how to develop data characters with common data story plots. With this work, we present a framework for data characters derived from our analysis; we then offer our extension to the data storytelling process using character-oriented design. To access our supplemental materials please visit https://chaorientdesignds.github.io/.

A common way to evaluate the reliability of dimensionality reduction (DR) embeddings is to quantify how well labeled classes form compact, mutually separated clusters in the embeddings. This approach is based on the assumption that the classes stay as clear clusters in the original high-dimensional space. However, in reality, this assumption can be violated; a single class can be fragmented into multiple separated clusters, and multiple classes can be merged into a single cluster. We thus cannot always assure the credibility of the evaluation using class labels. In this paper, we introduce two novel quality measures-Label-Trustworthiness and Label-Continuity (Label-T&C)-advancing the process of DR evaluation based on class labels. Instead of assuming that classes are well-clustered in the original space, Label-T&C work by (1) estimating the extent to which classes form clusters in the original and embedded spaces and (2) evaluating the difference between the two. A quantitative evaluation showed that Label-T&C outperform widely used DR evaluation measures (e.g., Trustworthiness and Continuity, Kullback-Leibler divergence) in terms of the accuracy in assessing how well DR embeddings preserve the cluster structure, and are also scalable. Moreover, we present case studies demonstrating that Label-T&C can be successfully used for revealing the intrinsic characteristics of DR techniques and their hyperparameters.

Recent advancements in pre-trained language-image models have ushered in a new era of visual comprehension. Leveraging the power of these models, this article tackles two issues within the realm of visual analytics: (1) the efficient exploration of large-scale image datasets and identification of data biases within them; (2) the evaluation of image captions and steering of their generation process. On the one hand, by visually examining the captions generated from language-image models for an image dataset, we gain deeper insights into the visual contents, unearthing data biases that may be entrenched within the dataset. On the other hand, by depicting the association between visual features and textual captions, we expose the weaknesses of pre-trained language-image models in their captioning capability and propose an interactive interface to steer caption generation. The two parts have been coalesced into a coordinated visual analytics system, fostering the mutual enrichment of visual and textual contents. We validate the effectiveness of the system with domain practitioners through concrete case studies with large-scale image datasets.

Implicit neural networks have demonstrated immense potential in compressing volume data for visualization. However, despite their advantages, the high costs of training and inference have thus far limited their application to offline data processing and non-interactive rendering. In this article, we present a novel solution that leverages modern GPU tensor cores, a well-implemented CUDA machine learning framework, an optimized global-illumination-capable volume rendering algorithm, and a suitable acceleration data structure to enable real-time direct ray tracing of volumetric neural representations. Our approach produces high-fidelity neural representations with a peak signal-to-noise ratio (PSNR) exceeding 30 dB, while reducing their size by up to three orders of magnitude. Remarkably, we show that the entire training step can fit within a rendering loop, bypassing the need for pre-training. Additionally, we introduce an efficient out-of-core training strategy to support extreme-scale volume data, making it possible for our volumetric neural representation training to scale up to terascale on a workstation with an NVIDIA RTX 3090 GPU. Our method significantly outperforms state-of-the-art techniques in terms of training time, reconstruction quality, and rendering performance, making it an ideal choice for applications where fast and accurate visualization of large-scale volume data is paramount.

Despite the value of VR (Virtual Reality) for educational purposes, the instructional power of VR in Biology Laboratory education remains under-explored. Laboratory lectures can be challenging due to students’ low motivation to learn abstract scientific concepts and low retention rate. Therefore, we designed a VR-based lecture on fermentation and compared its effectiveness with lectures using PowerPoint slides and a desktop application. Grounded in the theory of distributed cognition and motivational theories, our study examined how learning happens in each condition from students’ learning outcomes, behaviors, and perceptions. Our result indicates that VR facilitates students’ long-term retention to learn by cultivating their longer visual attention and fostering a higher sense of immersion, though students’ short-term retention remains the same across all conditions. This study extends current research on VR studies by identifying the characteristics of each teaching artifact and providing design implications for integrating VR technology into higher education.

With the most advanced natural language processing and artificial intelligence approaches, effective summarization of long and multi-topic documents—such as academic papers—for readers from different domains still remains a challenge. To address this, we introduce ConceptEVA, a mixed-initiative approach to generate, evaluate, and customize summaries for long and multi-topic documents. ConceptEVA incorporates a custom multi-task longformer encoder decoder to summarize longer documents. Interactive visualizations of document concepts as a network reflecting both semantic relatedness and co-occurrence help users focus on concepts of interest. The user can select these concepts and automatically update the summary to emphasize them. We present two iterations of ConceptEVA evaluated through an expert review and a within-subjects study. We find that participants’ satisfaction with customized summaries through ConceptEVA is higher than their own manually-generated summary, while incorporating critique into the summaries proved challenging. Based on our findings, we make recommendations for designing summarization systems incorporating mixed-initiative interactions.

Dimensionality reduction (DR) plays a vital role in the visual analysis of high-dimensional data. One main aim of DR is to reveal hidden patterns that lie on intrinsic low-dimensional manifolds. However, DR often overlooks important patterns when the manifolds are distorted or masked by certain influential data attributes. This paper presents a feature learning framework, FEALM, designed to generate a set of optimized data projections for nonlinear DR in order to capture important patterns in the hidden manifolds. These projections produce maximally different nearest-neighbor graphs so that resultant DR outcomes are significantly different. To achieve such a capability, we design an optimization algorithm as well as introduce a new graph dissimilarity measure, named neighbor-shape dissimilarity. Additionally, we develop interactive visualizations to assist comparison of obtained DR results and interpretation of each DR result. We demonstrate FEALM’s effectiveness through experiments and case studies using synthetic and real-world datasets.

Animal infectious diseases interfere with the sustainability of livestock farming. Developing comprehensive strategies for disease prevention and control requires professionals to study livestock farms from a variety of data sources, such as veterinary medical tests, financial reports, and animal movements between farms. However, investigating animal health surveillance is challenging as the collected data is often heterogeneous, high-dimensional, and spatio-temporal. Furthermore, data missingness, one common challenge in disease surveillance, can limit the effectiveness of the analysis and induce the misinterpretation of the result due to the lack of uncertainty representation. In this paper, we present a visual analytics interface of coordinated views that supports investigating disease outbreaks by connecting the relationships of livestock farms from different aspects — geospatial, transactional, and financial. Coupled with unsupervised machine learning methods, we infer the health status of a farm, despite the absence of its diagnostic history, with uncertainty and provide interpretability to such inferences. With these functionalities, we further quantify the influence of a disease outbreak, severity and scale, guiding the user toward investigating important outbreaks. We demonstrate the analysis capability of our visual analytics interface with multiple use cases on a real-world swine production dataset.

With the rapid accumulation of text data produced by data-driven techniques, the task of extracting "data annotations"—concise, high-quality data summaries from unstructured raw text—has become increasingly important. The recent advances in weak supervision and crowd-sourcing techniques provide promising solutions to efficiently create annotations (labels) for large-scale technical text data. However, such annotations may fail in practice because of the change in annotation requirements, application scenarios, and modeling goals, where label validation and relabeling by domain experts are required. To approach this issue, we present LabelVizier, a human-in-the-loop workflow that incorporates domain knowledge and user-specific requirements to reveal actionable insights into annotation flaws, then produce better-quality labels for large-scale multi-label datasets. We implement our workflow as an interactive notebook to facilitate flexible error profiling, in-depth annotation validation for three error types, and efficient annotation relabeling on different data scales. We evaluated our workflow in assisting the validation and relabelling of technical text annotation with two use cases and four expert reviews. The results show that LabelVizier is applicable in various application scenarios, and users with different knowledge backgrounds have diverse preferences for the tool usage.

Abstract A common way to render cell‐centric adaptive mesh refinement (AMR) data is to compute the dual mesh and visualize that with a standard unstructured element renderer. While the dual mesh provides a high‐quality interpolator, the memory requirements of the dual meshdata structureare significantly higher than those of the original grid, which prevents rendering very large data sets. We introduce a GPU‐friendly data structure and a clustering algorithm that allow for efficient AMR dual mesh rendering with a competitive memory footprint. Fundamentally, any off‐the‐shelf unstructured element renderer running on GPUs could be extended to support our data structure just by adding agridletelement type in addition to the standard tetrahedra, pyramids, wedges, and hexahedra supported by default. We integrated the data structure into a volumetric path tracer to compare it to various state‐of‐the‐art unstructured element sampling methods. We show that our data structure easily competes with these methods in terms of rendering performance, but is much more memory‐efficient.

We present an approach that shows all relevant subspaces of categorical data condensed in a single picture. We model the categorical values of the attributes as co-occurrences with data partitions generated from structured data using pattern mining. We show that these co-occurrences are a-priori , allowing us to greatly reduce the search space, effectively generating the condensed picture where conventional approaches filter out several subspaces as these are deemed insignificant. The task of identifying interesting subspaces is common but difficult due to exponential search spaces and the curse of dimensionality. One application of such a task might be identifying a cohort of patients defined by attributes such as gender, age, and diabetes type that share a common patient history, which is modeled as event sequences. Filtering the data by these attributes is common but cumbersome and often does not allow a comparison of subspaces. We contribute a powerful multi-dimensional pattern exploration approach (MDPE-approach) agnostic to the structured data type that models multiple attributes and their characteristics as co-occurrences, allowing the user to identify and compare thousands of subspaces of interest in a single picture. In our MDPE-approach, we introduce two methods to dramatically reduce the search space, outputting only the boundaries of the search space in the form of two tables. We implement the MDPE-approach in an interactive visual interface (MDPE-vis) that provides a scalable, pixel-based visualization design allowing the identification, comparison, and sense-making of subspaces in structured data. Our case studies using a gold-standard dataset and external domain experts confirm our approach’s and implementation’s applicability. A third use case sheds light on the scalability of our approach and a user study with 15 participants underlines its usefulness and power.

Adversarial attacks on a convolutional neural network (CNN)—injecting human-imperceptible perturbations into an input image—could fool a high-performance CNN into making incorrect predictions. The success of adversarial attacks raises serious concerns about the robustness of CNNs, and prevents them from being used in safety-critical applications, such as medical diagnosis and autonomous driving. Our work introduces a visual analytics approach to understanding adversarial attacks by answering two questions: (1) Which neurons are more vulnerable to attacks? and (2) Which image features do these vulnerable neurons capture during the prediction? For the first question, we introduce multiple perturbation-based measures to break down the attacking magnitude into individual CNN neurons and rank the neurons by their vulnerability levels. For the second, we identify image features (e.g., cat ears) that highly stimulate a user-selected neuron to augment and validate the neuron’s responsibility. Furthermore, we support an interactive exploration of a large number of neurons by aiding with hierarchical clustering based on the neurons’ roles in the prediction. To this end, a visual analytics system is designed to incorporate visual reasoning for interpreting adversarial attacks. We validate the effectiveness of our system through multiple case studies as well as feedback from domain experts.

Volume data is found in many important scientific and engineering applications. Rendering this data for visualization at high quality and interactive rates for demanding applications such as virtual reality is still not easily achievable even using professional-grade hardware. We introduce FoVolNet-a method to significantly increase the performance of volume data visualization. We develop a cost-effective foveated rendering pipeline that sparsely samples a volume around a focal point and reconstructs the full-frame using a deep neural network. Foveated rendering is a technique that prioritizes rendering computations around the user's focal point. This approach leverages properties of the human visual system, thereby saving computational resources when rendering data in the periphery of the user's field of vision. Our reconstruction network combines direct and kernel prediction methods to produce fast, stable, and perceptually convincing output. With a slim design and the use of quantization, our method outperforms state-of-the-art neural reconstruction techniques in both end-to-end frame times and visual quality. We conduct extensive evaluations of the system's rendering performance, inference speed, and perceptual properties, and we provide comparisons to competing neural image reconstruction techniques. Our test results show that FoVolNet consistently achieves significant time saving over conventional rendering while preserving perceptual quality.

Optimizing the performance of large-scale parallel codes is critical for efficient utilization of computing resources. Code developers often explore various execution parameters, such as hardware configurations, system software choices, and application parameters, and are interested in detecting and understanding bottlenecks in different executions. They often collect hierarchical performance profiles represented as call graphs, which combine performance metrics with their execution contexts. The crucial task of exploring multiple call graphs together is tedious and challenging because of the many structural differences in the execution contexts and significant variability in the collected performance metrics (e.g., execution runtime). In this paper, we present Ensemble CallFlow to support the exploration of ensembles of call graphs using new types of visualizations, analysis, graph operations, and features. We introduce ensemble-Sankey, a new visual design that combines the strengths of resource-flow (Sankey) and box-plot visualization techniques. Whereas the resource-flow visualization can easily and intuitively describe the graphical nature of the call graph, the box plots overlaid on the nodes of Sankey convey the performance variability within the ensemble. Our interactive visual interface provides linked views to help explore ensembles of call graphs, e.g., by facilitating the analysis of structural differences, and identifying similar or distinct call graphs. We demonstrate the effectiveness and usefulness of our design through case studies on large-scale parallel codes.

Depending on the node ordering, an adjacency matrix can highlight distinct characteristics of a graph. Deriving a "proper" node ordering is thus a critical step in visualizing a graph as an adjacency matrix. Users often try multiple matrix reorderings using different methods until they find one that meets the analysis goal. However, this trial-and-error approach is laborious and disorganized, which is especially challenging for novices. This paper presents a technique that enables users to effortlessly find a matrix reordering they want. Specifically, we design a generative model that learns a latent space of diverse matrix reorderings of the given graph. We also construct an intuitive user interface from the learned latent space by creating a map of various matrix reorderings. We demonstrate our approach through quantitative and qualitative evaluations of the generated reorderings and learned latent spaces. The results show that our model is capable of learning a latent space of diverse matrix reorderings. Most existing research in this area generally focused on developing algorithms that can compute "better" matrix reorderings for particular circumstances. This paper introduces a fundamentally new approach to matrix visualization of a graph, where a machine learning model learns to generate diverse matrix reorderings of a graph.

Vision transformer (ViT) expands the success of transformer models from sequential data to images. The model decomposes an image into many smaller patches and arranges them into a sequence. Multi-head self-attentions are then applied to the sequence to learn the attention between patches. Despite many successful interpretations of transformers on sequential data, little effort has been devoted to the interpretation of ViTs, and many questions remain unanswered. For example, among the numerous attention heads, which one is more important? How strong are individual patches attending to their spatial neighbors in different heads? What attention patterns have individual heads learned? In this work, we answer these questions through a visual analytics approach. Specifically, we first identify what heads are more important in ViTs by introducing multiple pruning-based metrics. Then, we profile the spatial distribution of attention strengths between patches inside individual heads, as well as the trend of attention strengths across attention layers. Third, using an autoencoder-based learning solution, we summarize all possible attention patterns that individual heads could learn. Examining the attention strengths and patterns of the important heads, we answer why they are important. Through concrete case studies with experienced deep learning experts on multiple ViTs, we validate the effectiveness of our solution that deepens the understanding of ViTs from head importance, head attention strength, and head attention pattern.

Spatial statistical analysis of multivariate volumetric data can be challenging due to scale, complexity, and occlusion. Advances in topological segmentation, feature extraction, and statistical summarization have helped overcome the challenges. This work introduces a new spatial statistical decomposition method based on level sets, connected components, and a novel variation of the restricted centroidal Voronoi tessellation that is better suited for spatial statistical decomposition and parallel efficiency. The resulting data structures organize features into a coherent nested hierarchy to support flexible and efficient out-of-core region-of-interest extraction. Next, we provide an efficient parallel implementation. Finally, an interactive visualization system based on this approach is designed and then applied to turbulent combustion data. The combined approach enables an interactive spatial statistical analysis workflow for large-scale data with a top-down approach through multiple-levels-of-detail that links phase space statistics with spatial features.

Environmental sensors provide crucial data for understanding our surroundings. For example, air quality maps based on sensor readings help users make decisions to mitigate the effects of pollution on their health. Standard maps show readings from individual sensors or colored contours indicating estimated pollution levels. However, showing a single estimate may conceal uncertainty and lead to underestimation of risk, while showing sensor data yields varied interpretations. We present several visualizations of uncertainty in air quality maps, including a frequency-framing "dotmap" and small multiples, and we compare them with standard contour and sensor-based maps. In a user study, we find that including uncertainty in maps has a significant effect on how much users would choose to reduce physical activity, and that people make more cautious decisions when using uncertainty-aware maps. Additionally, we analyze think-aloud transcriptions from the experiment to understand more about how the representation of uncertainty influences people's decision-making. Our results suggest ways to design maps of sensor data that can encourage certain types of reasoning, yield more consistent responses, and convey risk better than standard maps.

Diffusion tensor imaging (DTI) has been used to study the effects of neurodegenerative diseases on neural pathways, which may lead to more reliable and early diagnosis of these diseases as well as a better understanding of how they affect the brain. We introduce a predictive visual analytics system for studying patient groups based on their labeled DTI fiber tract data and corresponding statistics. The system's machine-learning-augmented interface guides the user through an organized and holistic analysis space, including the statistical feature space, the physical space, and the space of patients over different groups. We use a custom machine learning pipeline to help narrow down this large analysis space and then explore it pragmatically through a range of linked visualizations. We conduct several case studies using DTI and T1-weighted images from the research database of Parkinson's Progression Markers Initiative.

Real-world machine learning applications need to be thoroughly evaluated to meet critical product requirements for model release, to ensure fairness for different groups or individuals, and to achieve a consistent performance in various scenarios. For example, in autonomous driving, an object classification model should achieve high detection rates under different conditions of weather, distance, etc. Similarly, in the financial setting, credit-scoring models must not discriminate against minority groups. These conditions or groups are called as "Data Slices". In product MLOps cycles, product developers must identify such critical data slices and adapt models to mitigate data slice problems. Discovering where models fail, understanding why they fail, and mitigating these problems, are therefore essential tasks in the MLOps life-cycle. In this paper, we present SliceTeller, a novel tool that allows users to debug, compare and improve machine learning models driven by critical data slices. SliceTeller automatically discovers problematic slices in the data, helps the user understand why models fail. More importantly, we present an efficient algorithm, SliceBoosting, to estimate trade-offs when prioritizing the optimization over certain slices. Furthermore, our system empowers model developers to compare and analyze different model versions during model iterations, allowing them to choose the model version best suitable for their applications. We evaluate our system with three use cases, including two real-world use cases of product development, to demonstrate the power of SliceTeller in the debugging and improvement of product-quality ML models.

Visual data storytelling is gaining importance as a means of presenting data-driven information or analysis results, especially to the general public. This has resulted in design principles being proposed for data-driven storytelling, and new authoring tools being created to aid such storytelling. However, data analysts typically lack sufficient background in design and storytelling to make effective use of these principles and authoring tools. To assist this process, we present ChartStory for crafting data stories from a collection of user-created charts, using a style akin to comic panels to imply the underlying sequence and logic of data-driven narratives. Our approach is to operationalize established design principles into an advanced pipeline that characterizes charts by their properties and similarities to each other, and recommends ways to partition, layout, and caption story pieces to serve a narrative. ChartStory also augments this pipeline with intuitive user interactions for visual refinement of generated data comics. We extensively and holistically evaluate ChartStory via a trio of studies. We first assess how the tool supports data comic creation in comparison to a manual baseline tool. Data comics from this study are subsequently compared and evaluated to ChartStory's automated recommendations by a team of narrative visualization practitioners. This is followed by a pair of interview studies with data scientists using their own datasets and charts who provide an additional assessment of the system. We find that ChartStory provides cogent recommendations for narrative generation, resulting in data comics that compare favorably to manually-created ones.

Analyzing air pollution data is challenging as there are various analysis focuses from different aspects: feature (what), space (where), and time (when). As in most geospatial analysis problems, besides high-dimensional features, the temporal and spatial dependencies of air pollution induce the complexity of performing analysis. Machine learning methods, such as dimensionality reduction, can extract and summarize important information of the data to lift the burden of understanding such a complicated environment. In this paper, we present a methodology that utilizes multiple machine learning methods to uniformly explore these aspects. With this methodology, we develop a visual analytic system that supports a flexible analysis workflow, allowing domain experts to freely explore different aspects based on their analysis needs. We demonstrate the capability of our system and analysis workflow supporting a variety of analysis tasks with multiple use cases.

To maintain a robust and reliable supercomputing facility, monitoring it and understanding its hardware system events and behaviors is an essential task. Exascale systems will be increasingly heterogeneous, and the volume of systems data, collected from multiple subsystems and components measured at multiple fidelity levels and temporal resolutions, will continue to grow. In this work, we aim to create an effective solution to analyze diverse and massive datasets gathered from the error logs, job logs, and environment logs of an HPC system, such as a Cray XC40 supercomputer. In this work, we build an end-to-end error log analysis system that analyzes the job logs and gleans insights from their correspondence with hardware error logs and environment logs despite their varying temporal and spatial resolutions. Our machine learning pipeline built in our system is ~92% accurate in predicting the job exit status and does so with sufficient lead time for evasive actions to be taken before the actual failure event occurs.

Finding the similarities and differences between groups of datasets is a fundamental analysis task. For high-dimensional data, dimensionality reduction (DR) methods are often used to find the characteristics of each group. However, existing DR methods provide limited capability and flexibility for such comparative analysis as each method is designed only for a narrow analysis target, such as identifying factors that most differentiate groups. This paper presents an interactive DR framework where we integrate our new DR method, called ULCA (unified linear comparative analysis), with an interactive visual interface. ULCA unifies two DR schemes, discriminant analysis and contrastive learning, to support various comparative analysis tasks. To provide flexibility for comparative analysis, we develop an optimization algorithm that enables analysts to interactively refine ULCA results. Additionally, the interactive visualization interface facilitates interpretation and refinement of the ULCA results. We evaluate ULCA and the optimization algorithm to show their efficiency as well as present multiple case studies using real-world datasets to demonstrate the usefulness of this framework.

Many real-world applications involve analyzing time-dependent phenomena, which are intrinsically functional, consisting of curves varying over a continuum (e.g., time). When analyzing continuous data, functional data analysis (FDA) provides substantial benefits, such as the ability to study the derivatives and to restrict the ordering of data. However, continuous data inherently has infinite dimensions, and for a long time series, FDA methods often suffer from high computational costs. The analysis problem becomes even more challenging when updating the FDA results for continuously arriving data. In this paper, we present a visual analytics approach for monitoring and reviewing time series data streamed from a hardware system with a focus on identifying outliers by using FDA. To perform FDA while addressing the computational problem, we introduce new incremental and progressive algorithms that promptly generate the magnitude-shape (MS) plot, which conveys both the functional magnitude and shape outlyingness of time series data. In addition, by using an MS plot in conjunction with an FDA version of principal component analysis, we enhance the analyst's ability to investigate the visually-identified outliers. We illustrate the effectiveness of our approach with two use scenarios using real-world datasets. The resulting tool is evaluated by industry experts using real-world streaming datasets.

Collecting and analyzing anonymous personal information is required as a part of data analysis processes, such as medical diagnosis and restaurant recommendation. Such data should ostensibly be stored so that specific individual information cannot be disclosed. Unfortunately, inference attacks-integrating background knowledge and intelligent models-hinder classic sanitization techniques like syntactic anonymity and differential privacy from exhaustively protecting sensitive information. As a solution, we introduce a three-stage approach empowered within a visual interface, which depicts underlying inference behaviors via a Bayesian Network and supports a customized defense against inference attacks from unknown adversaries. In particular, our approach visually explains the process details of the underlying privacy preserving models, allowing users to verify if the results sufficiently satisfy the requirements of privacy preservation. We demonstrate the effectiveness of our approach through two case studies and expert reviews.

The rapid development of Convolutional Neural Networks (CNNs) in recent years has triggered significant breakthroughs in many machine learning (ML) applications. The ability to understand and compare various CNN models available is thus essential. The conventional approach with visualizing each model's quantitative features, such as classification accuracy and computational complexity, is not sufficient for a deeper understanding and comparison of the behaviors of different models. Moreover, most of the existing tools for assessing CNN behaviors only support comparison between two models and lack the flexibility of customizing the analysis tasks according to user needs. This paper presents a visual analytics system, VAC-CNN (Visual Analytics for Comparing CNNs), that supports the in-depth inspection of a single CNN model as well as comparative studies of two or more models. The ability to compare a larger number of (e.g., tens of) models especially distinguishes our system from previous ones. With a carefully designed model visualization and explaining support, VAC-CNN facilitates a highly interactive workflow that promptly presents both quantitative and qualitative information at each analysis stage. We demonstrate VAC-CNN's effectiveness for assisting novice ML practitioners in evaluating and comparing multiple CNN models through two use cases and one preliminary evaluation study using the image classification tasks on the ImageNet dataset.

Data-driven scrollytelling has become a prevalent way of visual communication because of its comprehensive delivery of perspectives derived from the data. However, creating an expressive scrollytelling story requires both data and design literacy and is time-consuming. As a result, scrollytelling has been mainly used only by professional journalists to disseminate opinions. In this paper, we present an automatic method to generate expressive scrollytelling visualization, which can present easy-to-understand data facts through a carefully arranged sequence of views. The method first enumerates data facts of a given dataset, and scores and organizes them. The facts are further assembled, sequenced into a story, with reader input taken into consideration. Finally, visual graphs, transitions, and text descriptions are generated to synthesize the scrollytelling visualization. In this way, non-professionals can easily explore and share interesting perspectives from selected data attributes and fact types. We demonstrate the effectiveness and usability of our method through both use cases and an in-lab user study.

Analysis of large, high-dimensional, and heterogeneous datasets is challenging as no one technique is suitable for visualizing and clustering such data in order to make sense of the underlying information. For instance, heterogeneous logs detailing machine repair and maintenance in an organization often need to be analyzed to diagnose errors and identify abnormal patterns, formalize root-cause analyses, and plan preventive maintenance. Such real-world datasets are also beset by issues such as inconsistent and/or missing entries. To conduct an effective diagnosis, it is important to extract and understand patterns from the data with support from analytic algorithms (e.g., finding that certain kinds of machine complaints occur more in the summer) while involving the human-in-the-loop. To address these challenges, we adopt existing techniques for dimensionality reduction (DR) and clustering of numerical, categorical, and text data dimensions, and introduce a visual analytics approach that uses multiple coordinated views to connect DR + clustering results across each kind of the data dimension stated. To help analysts label the clusters, each clustering view is supplemented with techniques and visualizations that contrast a cluster of interest with the rest of the dataset. Our approach assists analysts to make sense of machine maintenance logs and their errors. Then the gained insights help them carry out preventive maintenance. We illustrate and evaluate our approach through use cases and expert studies respectively, and discuss generalization of the approach to other heterogeneous data.

Current text visualization techniques typically provide overviews of document content and structure using intrinsic properties such as term frequencies, co-occurrences, and sentence structures. Such visualizations lack conceptual overviews incorporating domain-relevant knowledge, needed when examining documents such as research articles or technical reports. To address this shortcoming, we present ConceptScope, a technique that utilizes a domain ontology to represent the conceptual relationships in a document in the form of a Bubble Treemap visualization. Multiple coordinated views of document structure and concept hierarchy with text overviews further aid document analysis. ConceptScope facilitates exploration and comparison of single and multiple documents respectively. We demonstrate ConceptScope by visualizing research articles and transcripts of technical presentations in computer science. In a comparative study with DocuBurst, a popular document visualization tool, ConceptScope was found to be more informative in exploring and comparing domain-specific documents, but less so when it came to documents that spanned multiple disciplines.

Eye-tracking enabled virtual reality (VR) headsets have recently become more widely available. This opens up opportunities to incorporate eye gaze interaction methods in VR applications. However, studies on the fatigue-induced performance fluctuations of these new input modalities are scarce and rarely provide a direct comparison with established interaction methods. We conduct a study to compare the selection-interaction performance between commonly used handheld motion control devices and emerging eye interaction technology in VR. We investigate each interaction’s unique fatigue progression pattern in study sessions with ten minutes of continuous engagement. The results support and extend previous findings regarding the progression of fatigue in eye-tracked interaction over prolonged periods. By directly comparing gaze-with motion-controlled interaction, we put the emerging eye-trackers into perspective with the state-of-the-art interaction method for immersive space. We then discuss potential implications for future extended reality (XR) interaction design based on our findings.

Visualization transforms large quantities of data into pictures in which relations, patterns, or trends of interest in the data reveal themselves to effectively guide the user in the data reasoning and discovery process. Visualization has become an essential tool in many areas of study that use a data-driven approach to problem solving and decision making. However, when the data is large relational or high-dimensional, it can take both novices and experts substantial effort to derive and interpret visualization results from the data. Following the resurgence of AI and machine learning technology in recent years, in the field of visualization, there is also the growing interest and opportunity in applying AI and machine learning to perform data transformation and to assist in the generation and interpretation of visualization, aiming to strike a balance between cost and performance. In this talk, I will present designs made by my group effectively making use of machine learning for general data visualization and analytics tasks [1, 2, 3, 4, 5, 6], resulting in better visualization interfaces into the data.

Dynamic networks-networks that change over time-can be categorized into two types: offline dynamic networks, where all states of the network are known, and online dynamic networks, where only the past states of the network are known. Research on staging animated transitions in dynamic networks has focused more on offline data, where rendering strategies can take into account past and future states of the network. Rendering online dynamic networks is a more challenging problem since it requires a balance between timeliness for monitoring tasks-so that the animations do not lag too far behind the events-and clarity for comprehension tasks-to minimize simultaneous changes that may be difficult to follow. To illustrate the challenges placed by these requirements, we explore three strategies to stage animations for online dynamic networks: time-based, event-based, and a new hybrid approach that we introduce by combining the advantages of the first two. We illustrate the advantages and disadvantages of each strategy in representing low- and high-throughput data and conduct a user study involving monitoring and comprehension of dynamic networks. We also conduct a follow-up, think-aloud study combining monitoring and comprehension with experts in dynamic network visualization. Our findings show that animation staging strategies that emphasize comprehension do better for participant response times and accuracy. However, the notion of "comprehension" is not always clear when it comes to complex changes in highly dynamic networks, requiring some iteration in staging that the hybrid approach affords. Based on our results, we make recommendations for balancing event-based and time-based parameters for our hybrid approach.

We examine the process of designing an exhibit to communicate scientific findings from a complex dataset and unfamiliar domain to the public in a science museum. Our exhibit sought to communicate new lessons based on scientific findings from the domain of metagenomics. This multi-user exhibit had three goals: (1) to inform the public about microbial communities and their daily cycles; (2) to link microbes' activity to the concept of gene expression; (3) and to highlight scientists' use of gene expression data to understand the role of microbes. To address these three goals, we derived visualization designs with three corresponding stories, each corresponding to a goal. We present three successive rounds of design and evaluation of our attempts to convey these goals. We could successfully present one story but had limited success with our second and third goals. This work presents a detailed account of an attempt to explain tightly coupled relationships through storytelling and animation in a multi-user, informal learning environment to a public with varying prior knowledge on the domain and identify lessons for future design.

Data-driven problem solving in many real-world applications involves analysis of time-dependent multivariate data, for which dimensionality reduction (DR) methods are often used to uncover the intrinsic structure and features of the data. However, DR is usually applied to a subset of data that is either single-time-point multivariate or univariate time-series, resulting in the need to manually examine and correlate the DR results out of different data subsets. When the number of dimensions is large either in terms of the number of time points or attributes, this manual task becomes too tedious and infeasible. In this paper, we present MulTiDR, a new DR framework that enables processing of time-dependent multivariate data as a whole to provide a comprehensive overview of the data. With the framework, we employ DR in two steps. When treating the instances, time points, and attributes of the data as a 3D array, the first DR step reduces the three axes of the array to two, and the second DR step visualizes the data in a lower-dimensional space. In addition, by coupling with a contrastive learning method and interactive visualizations, our framework enhances analysts' ability to interpret DR results. We demonstrate the effectiveness of our framework with four case studies using real-world datasets.

We present P6, a declarative language for building high performance visual analytics systems through its support for specifying and integrating machine learning and interactive visualization methods. As data analysis methods based on machine learning and artificial intelligence continue to advance, a visual analytics solution can leverage these methods for better exploiting large and complex data. However, integrating machine learning methods with interactive visual analysis is challenging. Existing declarative programming libraries and toolkits for visualization lack support for coupling machine learning methods. By providing a declarative language for visual analytics, P6 can empower more developers to create visual analytics applications that combine machine learning and visualization methods for data analysis and problem solving. Through a variety of example applications, we demonstrate P6's capabilities and show the benefits of using declarative specifications to build visual analytics systems. We also identify and discuss the research opportunities and challenges for declarative visual analytics.

Calling context trees (CCTs) couple performance metrics with call paths, helping understand the execution and performance of parallel programs. To identify performance bottlenecks, programmers and performance analysts visually explore CCTs to form and validate hypotheses regarding degraded performance. However, due to the complexity of parallel programs, existing visual representations do not scale to applications running on a large number of processors. We present CallFlow, an interactive visual analysis tool that provides a high-level overview of CCTs together with semantic refinement operations to progressively explore CCTs. Using a flow-based metaphor, we visualize a CCT by treating execution time as a resource spent during the call chain, and demonstrate the effectiveness of our design with case studies on large-scale, production simulation codes.

In this paper, we present a visualization system for users to study multivariate time series data. They first identify trends or anomalies from a global view and then examine details in a local view. Specifically, we train a neural network to project high-dimensional data to a two dimensional (2D) planar space while retaining global data distances. By aligning the 2D points with a predefined color map, high-dimensional data can be represented by colors. Because perceptual color differentiation may fail to reflect data distance, we optimize perceptual color differentiation on each map region by deformation. The region with large perceptual color differentiation will expand, whereas the region with small differentiation will shrink. Since colors do not occupy any space in visualization, we convey the overview of multivariate time series data by a calendar view. Cells in the view are color-coded to represent multivariate data at different time spans. Users can observe color changes over time to identify events of interest. Afterward, they study details of an event by examining parallel coordinate plots. Cells in the calendar view and the parallel coordinate plots are dynamically linked for users to obtain insights that are barely noticeable in large datasets. The experiment results, comparisons, conducted case studies, and the user study indicate that our visualization system is feasible and effective.

Understanding and tuning the performance of extreme-scale parallel computing systems demands a streaming approach due to the computational cost of applying offline algorithms to vast amounts of performance log data. Analyzing large streaming data is challenging because the rate of receiving data and limited time to comprehend data make it difficult for the analysts to sufficiently examine the data without missing important changes or patterns. To support streaming data analysis, we introduce a visual analytic framework comprising of three modules: data management, analysis, and interactive visualization. The data management module collects various computing and communication performance metrics from the monitored system using streaming data processing techniques and feeds the data to the other two modules. The analysis module automatically identifies important changes and patterns at the required latency. In particular, we introduce a set of online and progressive analysis methods for not only controlling the computational costs but also helping analysts better follow the critical aspects of the analysis results. Finally, the interactive visualization module provides the analysts with a coherent view of the changes and patterns in the continuously captured performance data. Through a multi-faceted case study on performance analysis of parallel discrete-event simulation, we demonstrate the effectiveness of our framework for identifying bottlenecks and locating outliers.

The visualization of a network influences the quality of the mental map that the viewer develops to understand the network. In this study, we investigate the effects of a 3D immersive visualization environment compared to a traditional 2D desktop environment on the comprehension of a network’s structure. We compare the two visualization environments using three tasks—interpreting network structure, memorizing a set of nodes, and identifying the structural changes—commonly used for evaluating the quality of a mental map in network visualization. The results show that participants were able to interpret network structure more accurately when viewing the network in an immersive environment, particularly for larger networks. However, we found that 2D visualizations performed better than immersive visualization for tasks that required spatial memory.

Mind mapping is a popular way to explore a design space in creative thinking exercises, allowing users to form associations between concepts. Yet, most existing digital tools for mind mapping focus on authoring and organization, with little support for addressing the challenges of mind mapping such as stagnation and design fixation. We present Spinneret, a functional approach to aid mind mapping by providing suggestions based on a knowledge graph. Spinneret uses biased random walks to explore the knowledge graph in the neighborhood of an existing concept node in the mind map, and provides "suggestions" for the user to add to the mind map. A comparative study with a baseline mind-mapping tool reveals that participants created more diverse and distinct concepts with Spinneret, and reported that the suggestions inspired them to think of ideas they would otherwise not have explored.

Abstract Analyzing large and complex datasets for critical decision making can benefit from a collective effort involving a team of analysts. However, insights and findings from different analysts are often incomplete, disconnected, or even conflicting. Most existing analysis tools lack proper support for examining and resolving the conflicts among the findings in order to consolidate the results of collaborative data analysis. In this paper, we present CoVA, a visual analytics system incorporating conflict detection and resolution for supporting asynchronous collaborative data analysis. By using a declarative visualization language and graph representation for managing insights and insight provenance, CoVA effectively leverages distributed revision control workflow from software engineering to automatically detect and properly resolve conflicts in collaborative analysis results. In addition, CoVA provides an effective visual interface for resolving conflicts as well as combining the analysis results. We conduct a user study to evaluate CoVA for collaborative data analysis. The results show that CoVA allows better understanding and use of the findings from different analysts.

Dimensionality reduction (DR) methods are commonly used for analyzing and visualizing multidimensional data. However, when data is a live streaming feed, conventional DR methods cannot be directly used because of their computational complexity and inability to preserve the projected data positions at previous time points. In addition, the problem becomes even more challenging when the dynamic data records have a varying number of dimensions as often found in real-world applications. This paper presents an incremental DR solution. We enhance an existing incremental PCA method in several ways to ensure its usability for visualizing streaming multidimensional data. First, we use geometric transformation and animation methods to help preserve a viewer's mental map when visualizing the incremental results. Second, to handle data dimension variants, we use an optimization method to estimate the projected data positions, and also convey the resulting uncertainty in the visualization. We demonstrate the effectiveness of our design with two case studies using real-world datasets.

Dimensionality reduction (DR) is frequently used for analyzing and visualizing high-dimensional data as it provides a good first glance of the data. However, to interpret the DR result for gaining useful insights from the data, it would take additional analysis effort such as identifying clusters and understanding their characteristics. While there are many automatic methods (e.g., density-based clustering methods) to identify clusters, effective methods for understanding a cluster's characteristics are still lacking. A cluster can be mostly characterized by its distribution of feature values. Reviewing the original feature values is not a straightforward task when the number of features is large. To address this challenge, we present a visual analytics method that effectively highlights the essential features of a cluster in a DR result. To extract the essential features, we introduce an enhanced usage of contrastive principal component analysis (cPCA). Our method, called ccPCA (contrasting clusters in PCA), can calculate each feature's relative contribution to the contrast between one cluster and other clusters. With ccPCA, we have created an interactive system including a scalable visualization of clusters' feature contributions. We demonstrate the effectiveness of our method and system with case studies using several publicly available datasets.

Different layouts can characterize different aspects of the same graph. Finding a "good" layout of a graph is thus an important task for graph visualization. In practice, users often visualize a graph in multiple layouts by using different methods and varying parameter settings until they find a layout that best suits the purpose of the visualization. However, this trial-and-error process is often haphazard and time-consuming. To provide users with an intuitive way to navigate the layout design space, we present a technique to systematically visualize a graph in diverse layouts using deep generative models. We design an encoder-decoder architecture to learn a model from a collection of example layouts, where the encoder represents training examples in a latent space and the decoder produces layouts from the latent space. In particular, we train the model to construct a two-dimensional latent space for users to easily explore and generate various layouts. We demonstrate our approach through quantitative and qualitative evaluations of the generated layouts. The results of our evaluations show that our model is capable of learning and generalizing abstract concepts of graph layouts, not just memorizing the training examples. In summary, this paper presents a fundamentally new approach to graph visualization where a machine learning model learns to visualize a graph from examples without manually-defined heuristics.

We present P5, a web-based visualization toolkit that combines declarative visualization grammar and GPU computing for progressive data analysis and visualization. To interactively analyze and explore big data, progressive analytics and visualization methods have recently emerged. Progressive visualizations of incrementally refining results have the advantages of allowing users to steer the analysis process and make early decisions. P5 leverages declarative grammar for specifying visualization designs and exploits GPU computing to accelerate progressive data processing and rendering. The declarative specifications can be modified during progressive processing to create different visualizations for analyzing the intermediate results. To enable user interactions for progressive data analysis, P5 utilizes the GPU to automatically aggregate and index data based on declarative interaction specifications to facilitate effective interactive visualization. We demonstrate the effectiveness and usefulness of P5 through a variety of example applications and several performance benchmark tests.

We present P4, an information visualization toolkit that combines declarative design specification and GPU computing for building high-performance interactive systems. Most of the existing information visualization toolkits do not harness the power of parallel processors in today's mainstream computers. P4 leverages GPU computing to accelerate both data processing and visualization rendering for interactive visualization applications. P4's programming interface offers a declarative visualization grammar for rapid specifications of data transformations, visual encodings, and interactions. By simplifying the development of GPU-accelerated visualization systems while supporting a high degree of flexibility and customization for design specification, P4 narrows the gap between expressiveness and scalability in information visualization toolkits. Through a range of examples and benchmark tests, we demonstrate that P4 provides high efficiency for creating interactive visualizations and offers drastic performance improvement over current state-of-the-art toolkits.

This study describes a detailed analysis of museum visitors' decoding process as they used a visualization designed to support exploration of a large, complex dataset. Quantitative and qualitative analyses revealed that it took, on average, 43 seconds for visitors to decode enough of the visualization to see patterns and relationships in the underlying data represented, and 54 seconds to arrive at their first correct data interpretation. Furthermore, visitors decoded throughout and not only upon initial use of the visualization. The study analyzed think-aloud data to identify issues visitors had mapping the visual representations to their intended referents, examine why they occurred, and consider if and how these decoding issues were resolved. The paper also describes how multiple visual encodings both helped and hindered decoding and concludes with implications on the design and adaptation of visualizations for informal science learning venues.

The development of usable visualization solutions is essential for ensuring both their adoption and effectiveness. User-centered design principles, which involve users throughout the entire development process, have been shown to be effective in numerous information visualization endeavors. We describe how we applied these principles in scientific visualization over a two year collaboration to develop a hybrid in situ/post hoc solution tailored towards combustion researcher needs. Furthermore, we examine the importance of user-centered design and lessons learned over the design process in an effort to aid others seeking to develop effective scientific visualization solutions.

We conduct an eye tracking study to investigate perception text-embellished narrative visualizations under different task conditions. Study stimuli are data visualizations embellished with text-based elements: annotations, captions, labels, and descriptive text. We consider three common viewing tasks that occur when these types of graphics are viewed: (1) simple observation, (2) active search to answer a query, and (3) information memorization for later recall. The overarching goal is to understand, at a perceptual level, if and how task affects how these visualizations are interacted with. By analyzing collected gaze data and conducting advanced semantic scanpath analysis, we find, at a high level, diverse patterns of gaze behavior: simple observation and information memorization lead to similar optical viewing strategies, while active search significantly diverges, both in regards to which areas of the visualization are focused upon and how often embellishments are interacted with. We discuss study outcomes in the context of embellishing visualizations with text for various usage scenarios.

Machine learning for data-driven diagnosis has been actively studied in medicine to provide better healthcare. Supporting analysis of a patient cohort similar to a patient under treatment is a key task for clinicians to make decisions with high confidence. However, such analysis is not straightforward due to the characteristics of medical records: high dimensionality, irregularity in time, and sparsity. To address this challenge, we introduce a method for similarity calculation of medical records. Our method employs event and sequence embeddings. While we use an autoencoder for the event embedding, we apply its variant with the self-attention mechanism for the sequence embedding. Moreover, in order to better handle the irregularity of data, we enhance the self-attention mechanism with consideration of different time intervals. We have developed a visual analytics system to support comparative studies of patient records. To make a comparison of sequences with different lengths easier, our system incorporates a sequence alignment method. Through its interactive interface, the user can quickly identify patients of interest and conveniently review both the temporal and multivariate aspects of the patient records. We demonstrate the effectiveness of our design and system with case studies using a real-world dataset from the neonatal intensive care unit of UC Davis.

There is a growing trend of applying machine learning methods to medical datasets in order to predict patients’ future status. Although some of these methods achieve high performance, challenges still exist in comparing and evaluating different models through their interpretable information. Such analytics can help clinicians improve evidence-based medical decision making. In this work, we develop a visual analytics system that compares multiple models’ prediction criteria and evaluates their consistency. With our system, users can generate knowledge on different models’ inner criteria and how confidently we can rely on each model’s prediction for a certain patient. Through a case study of a publicly available clinical dataset, we demonstrate the effectiveness of our visual analytics system to assist clinicians and researchers in comparing and quantitatively evaluating different machine learning methods.

The task of data visualization generally involves a design step, which requires the knowledge of the data domain and visualization methods to do well.Because of the immense space for design optimization, it can take both novices and experts a tremendous effort to derive desired visualization results from data for exploration or communication.Following the resurgence of artificial intelligence technology in recent years, in the field of visualization, there is the growing interest and opportunity in applying AI to perform data transformation and to assist the generation of visualization, aiming to strike a balance between cost and quality.The use of visualization to enhance AI is the other active line of research.The PacificVis 2020 Workshop on Visualization Meets AI aims at exploring this emerging area of research and practice by fostering communication between visualization researchers and practitioners.This issue of Visual Informatics features the six papers chosen by the Workshop.Yiran Li et al. introduce a visual analytics system for comparing tree-based machine learning methods with respect to the reliability and interpretability of their predictions on patient records. Piyush Chawla et al. develop a technique to analyze a CNN based

Wildfires cause immense costs to human life, property, and the environment. As the impact of climate change increases the frequency and severity of wildfires, a renewed effort to understand these phenomena and their catalysts has increased. In this paper, we introduce a system that couples multiple sources of data and visualization to enable analysts to study historical fire data. We show two use cases to demonstrate the effectiveness of our system.

The distance plot (or unthresholded recurrence plot) has been shown to be a useful tool for analyzing spatiotemporal patterns in high-dimensional phase space trajectories. We incorporate this technique into an interactive visualization with multiple linked phase plots, and extend the distance plot to also visualize marker particle weights from particle-in-cell (PIC) simulations together with the phase space trajectories. By linking the distance plot with phase plots, one can more easily investigate the spatiotemporal patterns, and by extending the plot to visualize particle weights in conjunction with the phase space trajectories, the visualization better supports the needs of domain experts studying particle-in-cell simulations. We demonstrate our resulting visualization design using particles from an XGC Tokamak fusion simulation.

Solving complex data analysis problems can often benefit a collaborative effort. For synchronous co-located collaboration, a well-recognized challenge is to deliver different contents to people with different privileges and different responsibilities. This challenge is becoming more obvious with the use of a shared display space such as a wall-size display. In particular, scenarios often arise that a privileged participant needs to access sensitive information that other participants are not permitted to view. This is nearly impossible to achieve with only a single display. As a result, it becomes clear that additional devices are needed to provide some of the participants the capability to access and manage certain information in a private space. In this work, we investigate incorporating optical see-through head-mounted displays (OST-HMDs) with a wall-size display to deliver sensitive information in a synchronous co-located, collaborative setting. With our prototype system, we conduct a user study to observe the collaboration styles under this unique setup. We also present the lessons learned by reflecting on the iterative design process of our prototype system.

Group Support Systems provide ways to review and edit shared content during meetings, but typically require participants to explicitly generate the content. Recent advances in speech-to-text conversion and language processing now make it possible to automatically record and review spoken information. We present the iterative design and evaluation of TalkTraces, a real-time visualization that helps teams identify themes in their discussions and obtain a sense of agenda items covered. We use topic modeling to identify themes within the discussions and word embeddings to compute the discussion "relatedness" to items in the meeting agenda. We evaluate TalkTraces iteratively: we first conduct a comparative between-groups study between two teams using TalkTraces and two teams using traditional notes, over four sessions. We translate the findings into changes in the interface, further evaluated by one team over four sessions. Based on our findings, we discuss design implications for real-time displays of discussion content.

Abstract The inconceivable ability and common practice to collect personal data as well as the power of data‐driven approaches to businesses, services and security nowadays also introduce significant privacy issues. There have been extensive studies on addressing privacy preserving problems in the data mining community but relatively few have provided supervised control over the anonymization process. Preserving both the value and privacy of the data is largely a non‐trivial task. We present the design and evaluation of a visual interface that assists users in employing commonly used data anonymization techniques for making privacy preserving visualizations. Specifically, we focus on event sequence data due to its vulnerability to privacy concerns. Our interface is designed for data owners to examine potential privacy issues, obfuscate information as suggested by the algorithm and fine‐tune the results per their discretion. Multiple use case scenarios demonstrate the utility of our design. A user study similarly investigates the effectiveness of the privacy preserving strategies. Our results show that using a visual‐based interface is effective for identifying potential privacy issues, for revealing underlying anonymization processes, and for allowing users to balance between data utility and privacy.

Abstract We introduce a visual analysis system with GPU acceleration techniques for large sets of trajectories from complex dynamical systems. The approach is based on an interactive Boolean combination of subsets into a Focus+Context phase‐space visualization. We achieve high performance through efficient bitwise algorithms utilizing runtime generated GPU shaders and kernels. This enables a higher level of interactivity for visualizing the large multivariate trajectory data. We explain how our design meets a set of carefully considered analysis requirements, provide performance results, and demonstrate utility through case studies with many‐particle simulation data from two application areas.

We present an algorithm for parallel volume rendering that is a hybrid between classical object order and image order techniques. The algorithm operates on unstructured grids (and structured ones), and thus can deal with block boundaries interleaving in complex ways. It also deals effectively with cases that are prone to load imbalance, i.e., cases where cell sizes differ dramatically, either because of the nature of the input data, or because of the effects of the camera transformation. The algorithm divides work over resources such that each phase of its processing is bounded in the amount of computation it can perform. We demonstrate its efficacy through a series of studies, varying over camera position, data set size, transfer function, image size, and processor count. At its biggest, our experiments scaled up to 8,192 processors and operated on data sets with more than one billion cells. In total, we find that our hybrid algorithm performs well in all cases. This is because our algorithm naturally adapts its computation based on workload, and can operate like either an object order technique or an image order technique in scenarios where those techniques are efficient.

Face sketch synthesis shows great applications in a lot of fields such as online entertainment and suspects identification. Existing face sketch synthesis methods learn the patch-wise sketch style from the training dataset containing photo-sketch pairs. These methods manipulate the whole process directly in the field of RGB space, which unavoidably results in unsmooth noises at patch boundaries. If denoising methods are used, the sketch edges would be blurred and face structures could not be restored. Recent researches of feature maps, which are the outputs of a certain neural network layer, have achieved great success in texture synthesis and artistic image generation. In this paper, we reformulate the face sketch synthesis problem into a neural network feature maps based optimization task. Our results accurately capture the sketch drawing style and make full use of the whole stylistic information hidden in the training dataset. Unlike former feature map based methods, we utilize the Enhanced 3D PatchMatch and cross-layer cost aggregation methods to obtain the target feature maps for the final results. Multiple experiments have shown that our approach imitates hand-drawn sketch style vividly, and has high-quality visual effects on CUHK, AR, XM2VTS and CUFSF face sketch datasets.

This paper presents a declarative grammar for conveniently and effectively specifying advanced volume visualizations. Existing methods for creating volume visualizations either lack the flexibility to specify sophisticated visualizations or are difficult to use for those unfamiliar with volume rendering implementation and parameterization. Our design provides the ability to quickly create expressive visualizations without knowledge of the volume rendering implementation. It attempts to capture aspects of those difficult but powerful methods while remaining flexible and easy to use. As a proof of concept, our current implementation of the grammar allows users to combine multiple data variables in various ways and define transfer functions for diverse input data. The grammar also has the ability to describe advanced shading effects and create animations. We demonstrate the power and flexibility of our approach using multiple practical volume visualizations.

Analyzing social networks reveals the relationships between individuals and groups in the data. However, such analysis can also lead to privacy exposure (whether intentionally or inadvertently): leaking the real-world identity of ostensibly anonymous individuals. Most sanitization strategies modify the graph's structure based on hypothesized tactics that an adversary would employ. While combining multiple anonymization schemes provides a more comprehensive privacy protection, deciding the appropriate set of techniques-along with evaluating how applying the strategies will affect the utility of the anonymized results-remains a significant challenge. To address this problem, we introduce GraphProtector, a visual interface that guides a user through a privacy preservation pipeline. GraphProtector enables multiple privacy protection schemes which can be simultaneously combined together as a hybrid approach. To demonstrate the effectiveness of GraphProtector, we report several case studies and feedback collected from interviews with expert users in various scenarios.

Studies on augmenting visualization with sound are typically based on the assumption that sound can be complementary and assist in data analysis tasks. While sound promotes a different sense of engagement than vision, we conjecture that by augmenting non-speech audio to a visualization can not only help enhance the users' perception of the data but also increase their engagement with the data exploration process. We have designed a preliminary user study to test users' performance and engagement while exploring in a data visualization system under two different settings: visual-only and audiovisual. For our study, we used basketball player movement data in a game and created an interactive visualization system with three linked views. We supplemented sound to the visualization to enhance the users' understanding of a team's offensive/defensive behavior. The results of our study suggest that we need to better understand the effect of sound choice and encoding before considering engagement. We also find that sound can be useful to draw novice users' attention to patterns or anomalies in the data. Finally, we propose follow-up studies with designs informed by the findings from this study.

Deep Recurrent Neural Network (RNN) has gained popularity in many sequence classification tasks. Beyond predicting a correct class for each data instance, data scientists also want to understand what differentiating factors in the data have contributed to the classification during the learning process. We present a visual analytics approach to facilitate this task by revealing the RNN attention for all data instances, their temporal positions in the sequences, and the attribution of variables at each value level. We demonstrate with real-world datasets that our approach can help data scientists to understand such dynamics in deep RNNs from the training results, hence guiding their modeling process.

Researchers such as sociologists create visualizations of multivariate node-link diagrams to present findings about the relationships in communities. Unfortunately, such visualizations can inadvertently expose the ostensibly private identities of the persons that make up the dataset. By purposely violating graph readability metrics for a small region of the graph, we conjecture that local, exposed privacy leaks may be perceptually masked from easy recognition. In particular, we consider three commonly known metrics-edge crossing, node clustering, and node-edge overlapping-as a strategy to hide leaks. We evaluate the effectiveness of violating these metrics by conducting a user study that measures subject performance at visually searching for and identifying a privacy leak. Results show that when more masking operations are applied, participants needed more time to locate the privacy leak, though exhaustive, brute force search can eventually find it. We suggest future directions on how perceptual masking can be a viable strategy, primarily where modifying the underlying network structure is unfeasible.

Abstract Many data problems in the real world are complex and require multiple analysts working together to uncover embedded insights by creating chart‐driven data stories. How, as a subsequent analysis step, do we interpret and learn from these collections of charts? We present Chart Constellations, a system to interactively support a single analyst in the review and analysis of data stories created by other collaborative analysts. Instead of iterating through the individual charts for each data story, the analyst can project, cluster, filter, and connect results from all users in a meta‐visualization approach. Constellations supports deriving summary insights about prior investigations and supports the exploration of new, unexplored regions in the dataset. To evaluate our system, we conduct a user study comparing it against data science notebooks. Results suggest that Constellations promotes the discovery of both broad and high‐level insights, including theme and trend analysis, subjective evaluation, and hypothesis generation.

Rendering techniques that increase realism in volume visualization help enhance perception of the 3D features in the volume data. While techniques focusing on high-quality global illumination have been extensively studied, few works handle the interaction of light with materials in the volume. Existing techniques for light-material interaction are limited in their ability to handle high-frequency real-world material data, and the current treatment of volume data poorly supports the correct integration of surface materials. In this paper, we introduce an alternative definition for the transfer function which supports surface-like behavior at the boundaries between volume components and volume-like behavior within. We show that this definition enables multi-material rendering with high-quality, real-world material data. We also show that this approach offers an efficient alternative to pre-integrated rendering through isosurface techniques. We introduce arbitrary spatially-varying materials to achieve better multi-material support for scanned volume data. Finally, we show that it is possible to map an arbitrary set of parameters directly to a material representation for the more intuitive creation of novel materials.

Using different methods for laying out a graph can lead to very different visual appearances, with which the viewer perceives different information. Selecting a "good" layout method is thus important for visualizing a graph. The selection can be highly subjective and dependent on the given task. A common approach to selecting a good layout is to use aesthetic criteria and visual inspection. However, fully calculating various layouts and their associated aesthetic metrics is computationally expensive. In this paper, we present a machine learning approach to large graph visualization based on computing the topological similarity of graphs using graph kernels. For a given graph, our approach can show what the graph would look like in different layouts and estimate their corresponding aesthetic metrics. An important contribution of our work is the development of a new framework to design graph kernels. Our experimental study shows that our estimation calculation is considerably faster than computing the actual layouts and their aesthetic metrics. Also, our graph kernels outperform the state-of-the-art ones in both time and accuracy. In addition, we conducted a user study to demonstrate that the topological similarity computed with our graph kernel matches perceptual similarity assessed by human users.

In team-based workplaces, reviewing and reflecting on the content from a previously held meeting can lead to better planning and preparation. However, ineffective meeting summaries can impair this process, especially when participants have difficulty remembering what was said and what its context was. To assist with this process, we introduce MeetingVis, a visual narrative-based approach to meeting summarization. MeetingVis is composed of two primary components: (1) a data pipeline that processes the spoken audio from a group discussion, and (2) a visual-based interface that efficiently displays the summarized content. To design MeetingVis, we create a taxonomy of relevant meeting data points, identifying salient elements to promote recall and reflection. These are mapped to an augmented storyline visualization, which combines the display of participant activities, topic evolutions, and task assignments. For evaluation, we conduct a qualitative user study with five groups. Feedback from the study indicates that MeetingVis effectively triggers the recall of subtle details from prior meetings: all study participants were able to remember new details, points, and tasks compared to an unaided, memory-only baseline. This visual-based approaches can also potentially enhance the productivity of both individuals and the whole team.

Sharing data for public usage requires sanitization to prevent sensitive information from leaking. Previous studies have presented methods for creating privacy preserving visualizations. However, few of them provide sufficient feedback to users on how much utility is reduced (or preserved) during such a process. To address this, we design a visual interface along with a data manipulation pipeline that allows users to gauge utility loss while interactively and iteratively handling privacy issues in their data. Widely known and discussed types of privacy models, i.e., syntactic anonymity and differential privacy, are integrated and compared under different use case scenarios. Case study results on a variety of examples demonstrate the effectiveness of our approach.

Large, complex networks are commonly found in many application domains, such as sociology, biology, and software engineering. Analyzing such networks can be a non-trivial task, as it often takes many interactions to derive a finding. It is thus beneficial to capture and summarize the important steps in an analysis. This provenance would then effectively support recalling, reusing, reproducing, and sharing the analysis process and results. However, the provenance of analyzing a large, complex network would often be a long interaction record. To automatically compose a concise visual summarization of network analysis provenance, we introduce a ranking model together with a reduction algorithm. The model identifies and orders important interactions used in the network analysis. Based on this model, our algorithm is able to minimize the provenance, while still preserving all the essential steps for recalling and sharing the analysis process and results. We create a prototype system demonstrating the effectiveness of our model and algorithm with two usage scenarios.

The overall efficiency of an extreme-scale supercomputer largely relies on the performance of its network interconnects. Several of the state of the art supercomputers use networks based on the increasingly popular Dragonfly topology. It is crucial to study the behavior and performance of different parallel applications running on Dragonfly networks in order to make optimal system configurations and design choices, such as job scheduling and routing strategies. However, in order to study these temporal network behavior, we would need a tool to analyze and correlate numerous sets of multivariate time-series data collected from the Dragonfly’s multi-level hierarchies. This paper presents such a tool–a visual analytics system–that uses the Dragonfly network to investigate the temporal behavior and optimize the communication performance of a supercomputer. We coupled interactive visualization with time-series analysis methods to help reveal hidden patterns in the network behavior with respect to different parallel applications and system configurations. Our system also provides multiple coordinated views for connecting behaviors observed at different levels of the network hierarchies, which effectively helps visual analysis tasks. We demonstrate the effectiveness of the system with a set of case studies. Our system and findings can not only help improve the communication performance of supercomputing applications, but also the network performance of next-generation supercomputers.

Analyzing social network data helps sociologists understand the behaviors of individuals and groups as well as the relationships between them. With additional ontology information, the semantics behind the network structure can be further explored. Unfortunately, creating network visualizations with these datasets for presentation can inadvertently expose the private and sensitive information of individuals that reside in the data. To deal with this problem, we generalize conventional data anonymization models (originally designed for relational data) and formally apply them in the context of privacy preserving ontological network visualization. We use these models to identify the privacy leaks that exist in a visualization, provide graph modification actions that remove and/or perceptually minimize the effect of the identified leaks, and discuss strategies for what types of privacy actions to choose depending on the context of the leaks. We implement an ontological visualization interface with associated privacy preserving operations, and demonstrate with two case studies using real-world datasets to show that our approach can identify and solve potential privacy issues while balancing overall graph readability and utility.

Neuroscientists study brain functional connectivity in order to obtain a deeper understanding of how the brain functions. Current studies are mainly based on analyzing the averaged brain connectivity of a group (or groups) due to the high complexity of the collected data in terms of dimensionality, variability, and volume. While it is more desirable for the researchers to explore the potential variability between individual subjects or groups, a data analysis solution meeting this need is absent. In this paper, we present the design and capabilities of such a visual analytics system, which enables neuroscientists to visually compare the differences of brain networks between individual subjects as well as group averages, to explore a large dataset and examine sub-groups of participants that may not have been expected a priori to be of interest, to review detailed information as needed, and to manipulate the data and views to fit their analytical needs with easy interactions. We demonstrate the utility and strengths of this system with case studies using a representative functional connectivity dataset.

As virtual reality (VR) hardware technology becomes more mature and affordable, it is timely to develop visualization applications making use of such technology. How to interact with data in an immersive 3D space is both an interesting and challenging problem, demanding more research investigations. In this paper, we present a gesture input system for graph visualization in a stereoscopic 3D space. We compare desktop mouse input with gesture input with bare hands for performing a set of tasks on graphs. Our study results indicate that users are able to effortlessly manipulate and analyze graphs using gesture input. Furthermore, the results also show that using gestures is more efficient when exploring the complicated graph.

Volume rendering is an effective method for visualizing 3D data. However, it's still difficult to obtain an effective image, especially when there are complicated structures which may cause underexposure problems. Adjusting light sources and parameters adds computational cost, without alleviating the underexposure phenomenon. This paper presents the novel idea of applying lightness anchoring theory for volume visualization enhancement. An anchoring hypothesis, the Highest-Luminance-As-White rule, is adjusted to adapt our volume rendered image. After employing the lightness anchored optimization, underexposed areas can be revealed while still preserving the local depth relationship.

High-radix, low-diameter, hierarchical networks based on the Dragonfly topology are common picks for building next generation HPC systems. However, effective tools are lacking for analyzing the network performance and exploring the design choices for such emerging networks at scale. In this paper, we present visual analytics methods that couple data aggregation techniques with interactive visualizations for analyzing large-scale Dragonfly networks. We create an interactive visual analytics system based on these techniques. To facilitate effective analysis and exploration of network behaviors, our system provides intuitive, scalable visualizations that can be customized to show various traffic characteristics and correlate between different performance metrics. Using high-fidelity network simulation and HPC applications communication traces, we demonstrate the usefulness of our system with several case studies on exploring network behaviors at scale with different workloads, routing strategies, and job placement policies. Our simulations and visualizations provide valuable insights for mitigating network congestion and inter-job interference.

HPC systems have shifted to burst buffer storage and high radix interconnect topologies in order to meet the challenges of large-scale, data-intensive scientific computing. Both of these technologies have been studied in detail independently, but the interaction between them is not well understood. I/O traffic and communication traffic from concurrently scheduled applications may interfere with each other in unexpected ways, and this behavior may vary considerably depending on resource allocation, scheduling, and routing policies. In this work, we analyze I/O and network traffic interference on burst-buffer-equipped dragonfly-based systems using the high-resolution packet-level simulations provided by the CODES storage and interconnect simulation framework. The analysis is performed using realistic I/O workload sizes, a variety of resource allocation and network routing strategies employed in production environments, and a dragonfly network configuration modeled after current vendor options. We analyze the impact of interference on both I/O and communication traffic. We observe that although average network packet latency is stable across a wide variety of configurations, the maximum network packet latency in the presence of concurrent I/O traffic is highly sensitive to subtle policy changes. Our simulations reveal a worst-case single packet latency of 4,700 times the average latency for sub-optimal configurations. While a topology-aware mapping of compute nodes to burst buffer storage nodes can minimize the variation in maximum packet latency, it can slow down the I/O traffic by creating contention on the burst buffer nodes. Overall, balancing I/O and network performance requires careful selection of routing, data placement, and job placement policies.

With the recent advances in computer-supported cooperative work systems and increasing popularization of speech-based interfaces, groupware attempting to emulate a knowledgeable participant in a collaborative environment is bound to become a reality in the near future. In this paper, we present IdeaWall, a real-time system that continuously extracts essential information from a verbal discussion and augments that information with web-search materials. IdeaWall provides combinatorial visual stimuli to the participants to facilitate their creative process. We develop three cognitive strategies, from which a prototype application with three display modes was designed, implemented, and evaluated. The results of the user study with twelve groups show that IdeaWall effectively presents visual cues to facilitate verbal creative collaboration for idea generation and sets the stage for future research on intelligent systems that assist collaborative work.

Photo composition is an important factor affecting the aesthetics in photography. However, it is a highly challenging task to model the aesthetic properties of good compositions due to the lack of globally applicable rules to the wide variety of photographic styles. Inspired by the thinking process of photo taking, we formulate the photo composition problem as a view finding process which successively examines pairs of views and determines their aesthetic preferences. We further exploit the rich professional photographs on the web to mine unlimited high-quality ranking samples and demonstrate that an aesthetics-aware deep ranking network can be trained without explicitly modeling any photographic rules. The resulting model is simple and effective in terms of its architectural design and data sampling method. It is also generic since it naturally learns any photographic rules implicitly encoded in professional photographs. The experiments show that the proposed view finding network achieves state-of-the-art performance with sliding window search strategy on two image cropping datasets.

Immersive, stereoscopic visualization enables scientists to better analyze structural and physical phenomena compared to traditional display mediums. Unfortunately, current head-mounted displays (HMDs) with the high rendering quality necessary for these complex datasets are prohibitively expensive, especially in educational settings where their high cost makes it impractical to buy several devices. To address this problem, we develop two tools: (1) An authoring tool allows domain scientists to generate a set of connected, 360° video paths for traversing between dimensional keyframes in the dataset. (2) A corresponding navigational interface is a video selection and playback tool that can be paired with a low-cost HMD to enable an interactive, non-linear, storytelling experience. We demonstrate the authoring tool's utility by conducting several case studies and assess the navigational interface with a usability study. Results show the potential of our approach in effectively expanding the accessibility of high-quality, immersive visualization to a wider audience using affordable HMDs.

In transportation studies, one fundamental problem is to analyze the departures and arrivals at locations in order to predict the travel demands for urban planning and traffic management. These movements can relate to many factors, e.g., activity distributions and household demographics. This paper presents how we use visualization to explore the relationship between people movements and activity distributions that are characterized by the points of interest (POIs). To effectively model and visualize such relationship, we introduce POI-mobility signature, a compact visual representation with two main components. 1) A mobility component to present major people movements information across temporal dimension. 2) A POI component to present the activity context over an area of interest in spatial domain. To derive the signature, we study assorted analytical tasks after discussing with transportation researchers, consider essential design principles, and apply the representation to study a real-world dataset, which is the massive public transportation data in Singapore with over 30 million trajectories and crowd-sourcing POIs retrieved from Foursquare. Finally, we conduct three case studies and interview three transportation experts to verify the efficacy of our method.

The papers in this special issue were presented at IEEE VIS 2016, held during October 23-28, 2016 in Baltimore, MD. VIS contains three conferences, held concurrently: the IEEE Visual Analytics Science and Technology Conference (IEEE VAST 2016), the IEEE Information Visualization Conference (IEEE InfoVis 2016), and the IEEE Scientific Visualization Conference (IEEE SciVis2016).

Rapid advances in biology demand new tools for more active research dissemination and engaged teaching. This paper presents Synteny Explorer, an interactive visualization application designed to let college students explore genome evolution of mammalian species. The tool visualizes synteny blocks: segments of homologous DNA shared between various extant species that can be traced back or reconstructed in extinct, ancestral species. We take a karyogram-based approach to create an interactive synteny visualization, leading to a more appealing and engaging design for undergraduate-level genome evolution education. For validation, we conduct three user studies: two focused studies on color and animation design choices and a larger study that performs overall system usability testing while comparing our karyogram-based designs with two more common genome mapping representations in an educational context. While existing views communicate the same information, study participants found the interactive, karyogram-based views much easier and likable to use. We additionally discuss feedback from biology and genomics faculty, who judge Synteny Explorer's fitness for use in classrooms.

Visualization is a powerful technique for analysis and communication of complex, multidimensional, and time-varying data. However, it can be difficult to manually synthesize a coherent narrative in a chart or graph due to the quantity of visualized attributes, a variety of salient features, and the awareness required to interpret points of interest (POls). We present Temporal Summary Images (TSIs) as an approach for both exploring this data and creating stories from it. As a visualization, a TSI is composed of three common components: (1) a temporal layout, (2) comic strip-style data snapshots, and (3) textual annotations. To augment user analysis and exploration, we have developed a number of interactive techniques that recommend relevant data features and design choices, including an automatic annotations workflow. As the analysis and visual design processes converge, the resultant image becomes appropriate for data storytelling. For validation, we use a prototype implementation for TSIs to conduct two case studies with large-scale, scientific simulation datasets.

Visualizing distributions from data samples as well as spatial and temporal trends of multiple variables is fundamental to analyzing the output of today's scientific simulations. However, traditional visualization techniques are often subject to a trade-off between visual clutter and loss of detail, especially in a large-scale setting. In this work, we extend the use of spatially organized histograms into a sophisticated visualization system that can more effectively study trends between multiple variables throughout a spatial domain. Furthermore, we exploit the use of isosurfaces to visualize time-varying trends found within histogram distributions. This technique is adapted into both an on-the-fly scheme as well as an in situ scheme to maintain real-time interactivity at a variety of data scales.

The use of large-scale scientific simulations that can represent physical systems using both particle and volume data simultaneously is gaining popularity as each of these reference frames has an inherent set of advantages when studying different phenomena. Furthermore, being able to study the dynamic evolution of these time varying data types is an integral part of nearly all scientific endeavors. However, the techniques available to scientists generally limit them to studying each reference frame separately making it difficult to draw connections between the two. In this work we present a novel method of feature exploration that can be used to investigate spatio-temporal patterns in both data types simultaneously. More specifically, we focus on how spatio-temporal subsets can be identified from both reference frames, and develop new ways of visually presenting the embedded information to a user in an intuitive manner. We demonstrate the effectiveness of our method using case studies of real world scientific datasets and illustrate the new types of exploration and analyses that can be achieved through this technique.

The Eulerian and Lagrangian reference frames each provide a unique perspective when studying and visualizing results from scientific systems. As a result, many large-scale simulations produce data in both formats, and analysis tasks that simultaneously utilize information from both representations are becoming increasingly popular. However, due to their fundamentally different nature, drawing correlations between these data formats is a computationally difficult task, especially in a large-scale setting. In this work, we present a new data representation which combines both reference frames into a joint Eulerian-Lagrangian format. By reorganizing Lagrangian information according to the Eulerian simulation grid into a "unit cell" based approach, we can provide an efficient out-of-core means of sampling, querying, and operating with both representations simultaneously. We also extend this design to generate multi-resolution subsets of the full data to suit the viewer's needs and provide a fast flow-aware trajectory construction scheme. We demonstrate the effectiveness of our method using three large-scale real world scientific datasets and provide insight into the types of performance gains that can be achieved.

In this paper, we propose a framework for automatically producing thumbnails from stereo image pairs. It has two components focusing respectively on stereo saliency detection and stereo thumbnail generation. The first component analyzes stereo saliency through various saliency stimuli, stereoscopic perception and the relevance between two stereo views. The second component uses stereo saliency to guide stereo thumbnail generation. We develop two types of thumbnail generation methods, both changing image size automatically. The first method is called content-persistent cropping (CPC), which aims at cropping stereo images for display devices with different aspect ratios while preserving as much content as possible. The second method is an object-aware cropping method (OAC) for generating the smallest possible thumbnail pair that retains the most important content only and facilitates quick visual exploration of a stereo image database. Quantitative and qualitative experimental evaluations demonstrate promising performance of our thumbnail generation methods in comparison to state-of-the-art algorithms.

Human’s daily movements exhibit high regularity in a space–time context that typically forms circadian rhythms. Understanding the rhythms for human daily movements is of high interest to a variety of parties from urban planners, transportation analysts, to business strategists. In this paper, we present an interactive visual analytics design for understanding and utilizing data collected from tracking human’s movements. The resulting system identifies and visually presents frequent human movement rhythms to support interactive exploration and analysis of the data over space and time. Case studies using real-world human movement data, including massive urban public transportation data in Singapore and the MIT reality mining dataset, and interviews with transportation researches were conducted to demonstrate the effectiveness and usefulness of our system.

This paper presents a comparative study on personal visualizations of bibliographic data. We consider three designs for egocentric visualization: node-link diagrams, adjacency matrices, and botanical trees to depict one's academic career in terms of his/her publication records. Case studies are conducted to compare the effectiveness of resulting visualizations for conveying particular aspect of a researcher's bibliographic records. Based on our study, we find that node-link diagrams are better at revealing the overall distribution of certain attributes; adjacency matrices can convey more information with less clutter; and botanical trees are visually attractive and provide the best at a glance characterization of the mapped data, but mapping data to tree features must be carefully done to derive expressive visualization.

While the assessment of using motion in visualizations has been polarized, we conjecture that motion may be more effective in comparative visualizations if applied properly, especially when dealing with large amounts of multi-dimensional data. We have designed visualizations to represent driver behaviors. A series of user studies have been conducted to verify if adding motion to the static visualization can help users make comparisons and separate drastically different behaviors. Results show that adding motion indeed leads to shorter completion time and less cognitive workload.

We share our challenges and lessons learned in designing our exhibit prototype that encourages museum visitors to learn about marine animal behaviors through interactive visualization and data exploration. Our intent is to have visitors draw comparisons between animal behaviors, similarly to how scientists would, to make insights and discoveries. In our efforts, we have designed a set of visual encodings around the Tagging of Pelagic Predator (TOPP) data set to create the appropriate abstractions of this rich and complex field data. We have incorporated Multiple External Representations (MERs) and tangible user interfaces (TUIs) to provide a complementary representation of the data and promote self-learning. Through the formative evaluation, we can identify a few strengths and weaknesses of our prototype design. Our evaluation results suggest that we are progressing in the right direction - we observed the public making some comparisons and inferences - but still require further design iterations to improve our visualization exhibit.

Cosmological simulations produce a multitude of data types whose large scale makes them difficult to thoroughly explore in an interactive setting. One aspect of particular interest to scientists is the evolution of groups of dark matter particles, or "halos," described by merger trees. However, in order to fully understand subtleties in the merger trees, other data types derived from the simulation must be incorporated as well. In this work, we develop a novel interactive linked-view visualization system that focuses on simultaneously exploring dark matter halos, their hierarchical evolution, corresponding particle data, and other quantitative information. We employ a parallel remote renderer and a local merger tree selection tool so that users can analyze large data sets interactively. This allows scientists to assess their simulation code, understand inconsistencies in extracted data, and intuitively understand simulation behavior on all scales. We demonstrate the effectiveness of our system through a set of case studies on large-scale cosmological data from the HACC (Hardware/Hybrid Accelerated Cosmology Code) simulation framework.

Academic publication archives often draw from numerous, heterogeneous sources, whose records can follow differing naming conventions. As such, ambiguity issues concerning authorship of scientific papers often arise, such as authors sharing similar names, the use of first names versus initials, or alternate name spellings for the same author. These ambiguities have plagued research on scientific collaboration and influence. Detecting and correcting these errors is important for maintaining the archive, as well as for ensuring correctness and reliability in any desired subsequent analysis. There are existing analytic methods designed to accomplish this with varying degrees of accuracy, but many of them require fine tuning or manual categorization. We have developed a visual analytics system to interactively control and apply several analytic name disambiguation algorithms in a finely controlled manner, and to present the results to the user for verification or correction. We demonstrate the efficacy of our system by using it to find and resolve ambiguities in authorship data collected from Cornell University Library's arXiv.org and the InfoVis 2004 contest dataset with improved accuracy and speed over existing approaches.

Two-tiered direct network topologies such as Dragonflies have been proposed for future post-petascale and exascale machines, since they provide a high-radix, low-diameter, fast interconnection network. Such topologies call for redesigningMPI collective communication algorithms in order to attain the best performance. Yet as increasingly more applications share a machine, it is not clear how these topology-aware algorithms will react to interference with concurrent jobs accessing the same network. In this paper, we study three topology-aware broadcast algorithms, including one designed by ourselves. We evaluate their performance through event-driven simulation for small-and large-sized broadcasts (in terms of both data size and number of processes). We study the effect of different routing mechanisms on the topology-aware collective algorithms, as well as their sensitivity to network contention with other jobs. Our results show that while topology-aware algorithms dramatically reduce link utilization, their advantage in terms of latency is more limited.

Image mosaicing involves splitting an input image into a set of tiles, then replacing each tile with another image from a large dataset so that, when viewed from a distance, the resulting image resembles the original. We present a new approach for generating image mosaics using variable sized tiles made up from patches taken from photographs, paintings and texture images. This is different from previous work, where either simple regular tiling or adaptive tiling based on variations of RGB color was used. We propose an adaptive tiling theme by means of region entropy. In order to avoid the mismatch in roughness between the sub-image in the tile region of the input image and tile images in the dataset that may arise in the previous RGB color based image descriptors, we introduce the region entropy into the image descriptor to achieve better matching in both color and roughness. We also propose a new metric to measure the quality of the image mosaic which takes both the similarity and the mutual information between the generated mosaics and input images into account. The final mosaic images in this work are obtained by optimizing an objective function based on this metric.

Abstract Designing introductory materials is extremely important when developing new information visualization techniques. All users, regardless of their domain knowledge, first must learn how to interpret the visually encoded information in order to infer knowledge from visualizations. Yet, despite its significance, there has been little research on how to design effective introductory materials for information visualization. This paper presents a study on the design of online guides that educate new users on how to utilize information visualizations, particularly focusing on the employment of exercise questions in the guides. We use two concepts from educational psychology, learning type (or learning style) and teaching method, to design four unique types of online guides. The effects of the guides are measured by comprehension tests of a large group of crowdsourced participants. The tests covered four visualization types (graph, scatter plot, storyline, and tree map) and a complete range of visual analytics tasks. Our statistical analyses indicate that online guides which employ active learning and the top‐down teaching method are the most effective. Our study provides quantitative insight into the use of exercise questions in online guides for information visualizations and will inspire further research on design considerations for other elements in introductory materials.

Abstract Existing real‐time volume rendering techniques which support global illumination are limited in modeling distinct realistic appearances for classified volume data, which is a desired capability in many fields of study for illustration and education. Directly extending the emission‐absorption volume integral with heterogeneous material shading becomes unaffordable for real‐time applications because the high‐frequency view‐dependent global lighting needs to be evaluated per sample along the volume integral. In this paper, we present a decoupled shading algorithm for multi‐material volume rendering that separates global incident lighting evaluation from per‐sample material shading under multiple light sources. We show how the incident lighting calculation can be optimized through a sparse volume integration method. The quality, performance and usefulness of our new multi‐material volume rendering method is demonstrated through several examples.

Information visualization has traditionally limited itself to 2D representations, primarily due to the prevalence of 2D displays and report formats. However, there has been a recent surge in popularity of consumer grade 3D displays and immersive head-mounted displays (HMDs). The ubiquity of such displays enables the possibility of immersive, stereoscopic visualization environments. While techniques that utilize such immersive environments have been explored extensively for spatial and scientific visualizations, contrastingly very little has been explored for information visualization. In this paper, we present our considerations of layout, rendering, and interaction methods for visualizing graphs in an immersive environment. We conducted a user study to evaluate our techniques compared to traditional 2D graph visualization. The results show that participants answered significantly faster with a fewer number of interactions using our techniques, especially for more difficult tasks. While the overall correctness rates are not significantly different, we found that participants gave significantly more correct answers using our techniques for larger graphs.

Cloud computing is an essential technology to Big Data analytics and services. A cloud computing system is often comprised of a large number of parallel computing and storage devices. Monitoring the usage and performance of such a system is important for efficient operations, maintenance, and security. Tracing every application on a large cloud system is untenable due to scale and privacy issues. But profile data can be collected relatively efficiently by regularly sampling the state of the system, including properties such as CPU load, memory usage, network usage, and others, creating a set of multivariate time series for each system. Adequate tools for studying such large-scale, multidimensional data are lacking. In this paper, we present a visual based analysis approach to understanding and analyzing the performance and behavior of cloud computing systems. Our design is based on similarity measures and a layout method to portray the behavior of each compute node over time. When visualizing a large number of behavioral lines together, distinct patterns often appear suggesting particular types of performance bottleneck. The resulting system provides multiple linked views, which allow the user to interactively explore the data by examining the data or a selected subset at different levels of detail. Our case studies, which use datasets collected from two different cloud systems, show that this visual based approach is effective in identifying trends and anomalies of the systems.

While virtual reality has been researched in many ways for spatial and scientific visualizations, comparatively little has been explored for visualizations of more abstract kinds of data. In particular, stereoscopic and VR environments for graph visualization have only been applied as limited extensions to standard 2D techniques (e.g. using stereoscopy for highlighting). In this work, we explore a new, immersive approach for graph visualization, designed specifically for virtual reality environments.

The benefits of using advanced illumination models in volume visualization have been demonstrated by many researchers. Interactive volume rendering incorporated with advanced lighting has been achieved with GPU acceleration for regular-grid volume data, making volume visualization even more appealing as a tool for 3D data exploration. This paper presents an interactive illumination strategy, which is specially designed and optimized for volume visualization of unstructured-grid data. The basis of the design is a partial differential equation based illumination model to simulate the light propagation, absorption, and scattering within the volumetric medium. In particular, a two-level scheme is introduced to overcome the challenges presented by unstructured grids. Test results show that the added illumination effects such as global shadowing and multiple scattering not only lead to more visually pleasing visualization, but also greatly enhance the perception of the depth information and complex spatial relationships for features of interest in the volume data. This volume visualization enhancement is introduced at a time when unstructured grids are becoming increasingly popular for a variety of scientific simulation applications.

Dimensionality Reduction (DR) is a crucial tool to facilitate high-dimensional data analysis. As the volume and the variety of features used to describe a phenomenon keeps increasing, DR has become not only desirable but paramount. However, DR can result in unreliable depictions of data. The uncertainties involved in DR may stem from the selection of methods, parameter configurations, and the constraints imposed by the user. To address these uncertainties, various means of DR quality assessment have been proposed in the literature. Nevertheless, how to optimize the trade-off between the quantification efficiency and accuracy is yet to be further studied. The purpose of this paper is to present a general technique, in the context of visual analytics, to support efficient uncertainty-aware high-dimensional data exploration. We model the uncertainty based on how well neighborhood geometries are preserved during DR. We employ approximated nearest neighbor (ANN) search algorithms to speed up the quantification process with marginal decrease in accuracy. We then visualize the quantified uncertainties in the form of augmented scatter plot. We test our technique with three real world datasets against several well-known DR techniques, and discuss possible underlying causes that lead to certain embedding patterns. Our results show that our approach is effective and beneficial for both DR assessment and user-centered data exploration.

Design methodologies for information visualizations are typically based on the assumption that the users will be fully engaged in the visual exploration of the displayed information. However, recent research suggests that there is an increasing diversity in how users engage with modern visualizations, and that the traditional design theories do not always satisfy the varied users needs. In this paper, we present a new design concept, engagement-versatile design, for visualizations that target users with a variety of engagement styles. Without losing generality, we demonstrate the feasibility of this concept through the designing of a system called Stock Lamp, an engagement-versatile visualization that helps users keep track of the stock market in real-time. This design process includes identifying different modes of engagement, deriving design implications from each engagement-mode, and applying them to the visualization's design. Our user study shows that Stock Lamp is able to consistently relay market information even when the users are multi-tasking. We believe this study establishes a new concept that promotes a systematic design approach that leverages both theoretical and empirical design methodologies for future visualization development.

OBJECTIVE: The aim of this study is to analyze and visualize the polymorbidity associated with chronic kidney disease (CKD). The study shows diseases associated with CKD before and after CKD diagnosis in a time-evolutionary type visualization. MATERIALS AND METHODS: Our sample data came from a population of one million individuals randomly selected from the Taiwan National Health Insurance Database, 1998 to 2011. From this group, those patients diagnosed with CKD were included in the analysis. We selected 11 of the most common diseases associated with CKD before its diagnosis and followed them until their death or up to 2011. We used a Sankey-style diagram, which quantifies and visualizes the transition between pre- and post-CKD states with various lines and widths. The line represents groups and the width of a line represents the number of patients transferred from one state to another. RESULTS: The patients were grouped according to their states: that is, diagnoses, hemodialysis/transplantation procedures, and events such as death. A Sankey diagram with basic zooming and planning functions was developed that temporally and qualitatively depicts they had amid change of comorbidities occurred in pre- and post-CKD states. DISCUSSION: This represents a novel visualization approach for temporal patterns of polymorbidities associated with any complex disease and its outcomes. The Sankey diagram is a promising method for visualizing complex diseases and exploring the effect of comorbidities on outcomes in a time-evolution style. CONCLUSIONS: This type of visualization may help clinicians foresee possible outcomes of complex diseases by considering comorbidities that the patients have developed.

This paper presents a novel framework for applying storyline visualizations to streaming data. The framework includes three components: a new data management scheme for processing and storing the incoming data, a layout construction algorithm specifically designed for incrementally generating storylines from streaming data, and a layout refinement algorithm for improving the legibility of the visualization. By dividing the layout computation to two separate components, one for constructing and another for refining, our framework effectively provides the users with the ability to follow and reason dynamic data. The evaluation studies of our storyline visualization framework demonstrate its efficacy to present streaming data as well as its superior performance over existing methods in terms of both computational efficiency and visual clarity.

Computing distance fields is fundamental to many scientific and engineering applications. Distance fields can be used to direct analysis and reduce data. In this paper, we present a highly scalable method for computing 3D distance fields on massively parallel distributed-memory machines. A new distributed spatial data structure, named parallel distance tree, is introduced to manage the level sets of data and facilitate surface tracking over time, resulting in significantly reduced computation and communication costs for calculating the distance to the surface of interest from any spatial locations. Our method supports several data types and distance metrics from real-world applications. We demonstrate its efficiency and scalability on state-of-the-art supercomputers using both large-scale volume datasets and surface models. We also demonstrate in-situ distance field computation on dynamic turbulent flame surfaces for a petascale combustion simulation. Our work greatly extends the usability of distance fields for demanding applications.

Research into social network analysis has shown that graph metrics, such as degree and closeness, are often used to summarize structural changes in a dynamic graph. However there have been few visual analytics approaches that have been proposed to help analysts study graph evolutions in the context of graph metrics. In this paper, we present a novel approach, called GraphFlow, to visualize dynamic graphs. In contrast to previous approaches that provide users with an animated visualization, GraphFlow offers a static flow visualization that summarizes the graph metrics of the entire graph and its evolution over time. Our solution supports the discovery of high-level patterns that are difficult to identify in an animation or in individual static representations. In addition, GraphFlow provides users with a set of interactions to create filtered views. These views allow users to investigate why a particular pattern has occurred. We showcase the versatility of GraphFlow using two different datasets and describe how it can help users gain insights into complex dynamic graphs.

The ability to analyze and assimilate Electronic Medical Records (EMR) has great value to physicians, clinical researchers, and medical policy makers. Current EMR systems do not provide adequate support for fully exploiting the data. The growing size, complexity, and accessibility of EMRs demand a new set of tools for extracting knowledge of interest from the data. This paper presents an interactive visual mining solution for cohort study of EMRs. The basis of our design is multidimensional, visual aggregation of the EMRs. The resulting visualizations can help uncover hidden structures in the data, compare different patient groups, determine critical factors to a particular disease, and help direct further analyses. We introduce and demonstrate our design with case studies using EMRs of 14,567 Chronic Kidney Disease (CKD) patients.

Computers make incredible amounts of information available at our fingertips. As computers become integral parts of our lives, we spend more time staring at computer monitor than ever before, sometimes with negative effects. One major concern is the increasing number of people suffering from Computer Vision Syndrome (CVS). CVS is caused by extensive use of computers, and its symptoms include eye fatigue, frequent headaches, dry eyes, and blurred vision. It is possible to partially alleviate CVS if we can remind users to blink more often. We present a prototype system that uses a camera to monitor a user's blink rate, and when the user has not blinked in a while, the system triggers a blink stimulus. We investigated four different types of eye-blink stimulus: screen blurring, screen flashing, border flashing, and pop-up notifications. Users also rated each stimulus type in terms of effectiveness, intrusiveness, and satisfaction. Results from our user studies show that our stimuli are effective in increasing user blink rate with screen blurring being the best.

Image/video matting is one of the key operations in many image/video editing applications. Although previous methods can generate high-quality matting results, their high computational cost in processing high-resolution image and video data often limits their usability. In this paper, we present a unified acceleration method for closed-form image and video matting using a hierarchical data structure, which achieves an excellent compromise between quality and speed. We first apply a Gaussian KD tree to adaptively cluster the input high-dimensional image and video feature space into a low-dimensional feature space. Then, we solve the affinity-weighted Laplacian alpha matting in the reduced feature space. The final matting results are derived using detail-aware alpha interpolation. Our algorithm can be fully parallelized by exploiting advanced graphics hardware, which can further accelerate the matting computation. Our method accelerates existing methods by at least an order of magnitude with good quality, and also greatly reduces the memory consumption. This acceleration strategy is also extended to support other affinity-based matting approaches, which makes it a more general accelerating framework for a variety of matting methods. Finally, we apply the presented method to accelerate image and video dehazing, and image shadow detection and removal.

Scatterplots are commonly used to visualize multidimensional data; however, 2D projections of data offer limited understanding of the high-dimensional interactions between data points. We introduce an interactive 3D extension of scatterplots called the Regression Cube (RC), which augments a 3D scatterplot with three facets on which the correlations between the two variables are revealed by sensitivity lines and sensitivity streamlines. The sensitivity visualization of local regression on the 2D projections provides insights about the shape of the data through its orientation and continuity cues. We also introduce a series of visual operations such as clustering, brushing, and selection supported in RC. By iteratively refining the selection of data points of interest, RC is able to reveal salient local correlation patterns that may otherwise remain hidden with a global analysis. We have demonstrated our system with two examples and a user-oriented evaluation, and we show how RCs enable interactive visual exploration of multidimensional datasets via a variety of classification and information retrieval tasks. A video demo of RC is available.

Scatterplots are widely used to visualize scatter dataset for exploring outliers, clusters, local trends, and correlations. Depicting multi-class scattered points within a single scatterplot view, however, may suffer from heavy overdraw, making it inefficient for data analysis. This paper presents a new visual abstraction scheme that employs a hierarchical multi-class sampling technique to show a feature-preserving simplification. To enhance the density contrast, the colors of multiple classes are optimized by taking the multi-class point distributions into account. We design a visual exploration system that supports visual inspection and quantitative analysis from different perspectives. We have applied our system to several challenging datasets, and the results demonstrate the efficiency of our approach.

We present a novel scheme for progressive rendering in interactive visualization. Static settings with respect to a certain image quality or frame rate are inherently incapable of delivering both high frame rates for rapid changes and high image quality for detailed investigation. Our novel technique flexibly adapts by steering the visualization process in three major degrees of freedom: when to terminate the refinement of a frame in the background and start a new one, when to display a frame currently computed, and how much resources to consume. We base these decisions on the correlation of the errors due to insufficient sampling and response delay, which we estimate separately using fast yet expressive heuristics. To automate the configuration of the steering behavior, we employ offline video quality analysis. We provide an efficient implementation of our scheme for the application of volume raycasting, featuring integrated GPU-accelerated image reconstruction and error estimation. Our implementation performs an integral handling of the changes due to camera transforms, transfer function adaptations, as well as the progression of the data to in time. Finally, the overall technique is evaluated with an expert study.

In this paper, we present a high quality and interactive method for volume rendering curvilinear-grid data sets. This method is based on a two-stage parallel transformation of the sample position into intermediate computational space then into texture space through the use of multiple 1 and 2D deformation textures using hardware acceleration. In this manner, it is possible to render many curvilinear-grid volume data sets at high quality and with a low memory footprint, while taking advantage of modern graphic hardware's tri-linear filtering for the data itself. We also extend our method to handle volume shading. Additionally, we present a comprehensive study and comparisons with previous works, we show improvements both in quality and performance using our technique on multiple curvilinear data sets.

Video synopsis aims at removing video's less important information, while preserving its key content for fast browsing, retrieving, or efficient storing. Previous video synopsis methods, including frame-based and object-based approaches that remove valueless whole frames or combine objects from time shots, cannot handle videos with redundancies existing in the movements of video object. In this paper, we present a novel part-based object movements synopsis method, which can effectively compress the redundant information of a moving video object and represent the synopsized object seamlessly. Our method works by part-based assembling and stitching. The object movement sequence is first divided into several part movement sequences. Then, we optimally assemble moving parts from different part sequences together to produce an initial synopsis result. The optimal assembling is formulated as a part movement assignment problem on a Markov Random Field (MRF), which guarantees the most important moving parts are selected while preserving both the spatial compatibility between assembled parts and the chronological order of parts. Finally, we present a non-linear spatiotemporal optimization formulation to stitch the assembled parts seamlessly, and achieve the final compact video object synopsis. The experiments on a variety of input video objects have demonstrated the effectiveness of the presented synopsis method.

Studying the dynamic evolution of time-varying volumetric data is essential in countless scientific endeavors. The ability to isolate and track features of interest allows domain scientists to better manage large complex datasets both in terms of visual understanding and computational efficiency. This work presents a new trajectory-based feature tracking technique for use in joint particle/volume datasets. While traditional feature tracking approaches generally require a high temporal resolution, this method utilizes the indexed trajectories of corresponding Lagrangian particle data to efficiently track features over large jumps in time. Such a technique is especially useful for situations where the volume dataset is either temporally sparse or too large to efficiently track a feature through all intermediate timesteps. In addition, this paper presents a few other applications of this approach, such as the ability to efficiently track the internal properties of volumetric features using variables from the particle data. We demonstrate the effectiveness of this technique using real world combustion and atmospheric datasets and compare it to existing tracking methods to justify its advantages and accuracy.

Contact diaries are interpersonal communication logs which are obtained in sociological and epidemiological studies. These logs can be used to study the social patterns of communities over a period of time. A dataset composed of diaries maps well to a set of one-tiered, categorical, independent and egocentric networks. This paper presents an interface for visualization and analysis of contact diaries datasets using an interactive radial mapping scheme, with case studies illustrating a standard workflow using the application. We facilitate individual diary analysis, multi-dataset comparison, and an overlay interface for investigating a set of many diaries in a singular space. With this interface, network researchers can utilize visualization to enhance their analysis of contact diaries.

Online bibliographic databases have become widely available and are important resources to scientific researchers. These databases store rich information and many evolve into digital libraries. Using a bibliographic database of a specific discipline, we can extract a co-authorship and citation network of individual professionals. This allows for the study of patterns in scholarly contributions as well as for the exploration of scientific disputes associated with an individuals career. We have designed a visualization tool, which we call PathWay, to discover and understand patterns and trends in the bibliographic data over a selected period of time. With PathWay, we conducted case studies on a bibliography of approximately 400,000 scientists in physics over a 26 year time period. In this paper, we show how PathWay can be used to characterize one's academic career path in terms of the publication record, conduct comparative studies that would be difficult to do with conventional search methods, and also provide a way to gain insight into the emergence and the career implications of the scientific disputes associated with publications.

With the advances in dual medical imaging, the requirements for multimodal and multifield volume visualization begin to emerge. One of the challenges in multimodal visualization is how to simplify the process of generating informative pictures from complementary data. In this paper we present an automatic technique that makes use of dual modality information, such as CT and PET, to produce effective focus+context volume visualization. With volume ray casting, per-ray visibility histograms summarize the contribution of samples along each ray to the final image. By quantifying visibility for the region of interest, indicated by the PET data, occluding tissues can be made just transparent enough to give a clear view of the features in that region while preserving some context. Unlike most previous methods relying on costly-preprocessing and tedious manual tuning, our technique achieves comparable and better results based on on-the-fly processing that still enables interactive visualization. Our work thus offers a powerful visualization technique for examining multimodal volume data. We demonstrate the technique with scenarios for the detection and diagnosis of cancer and other pathologies.

Large dynamic graphs occur in many fields. While overviews are often used to provide summaries of the overall structure of the graph, they become less useful as data size increases. Often analysts want to focus on a specific part of the data according to domain knowledge, which is best suited by a bottom-up approach. This paper presents an egocentric, bottom-up method to exploring a large dynamic network using a storyline representation to summarise localized behavior of the network over time.

There has been a long-standing question of whether sound or audio annotations may assist in data visualization tasks. This paper presents our study focusing on enhancing timeline/storyline like visualizations with audio annotations. Timeline visualizations are widely used to illustrate interactions among different entities over time. This type of visualizations facilitate reviewing important activities and their associations. For a long timeline, however, it is difficult for the viewer to remember all the essential information found in the visualization process. Since hearing is another primary human sense for perceiving information, we conjecture that adding audio annotations to selected sections of a timeline visualization can help improve the viewer's recall of important events. We have designed a user study based on augmenting storyline visualizations with verbal text, and tested subjects with three different settings: visual cues only, verbal text only, and both. While our test results do not give a strong indication of the clear advantage of adding verbal text, the lessons learned in our study suggest directions for further study and will help us and others design audio-augmented visualization systems.

Web applications are abundant and used daily by a variety of people. Planning a trip, for example, has become much easier with Web applications such as Google Maps. Nevertheless, to thoroughly design a personalized trip, the user often needs to manually search for additional information and mentally combine all the information to make many key decisions. To address the need for a comprehensive system, we have designed OnMyWay, a mash up system which effectively incorporates the ability to search points of interest and to assess essential information for designing personalized road trips. This paper presents our task-oriented design for presenting information and supporting user interactions applied to OnMyWay, and provides evaluations of the system's design and usability based on two informal user studies.

Parallel applications rely on I/O to load data, store end results, and protect partial results from being lost to system failure. Parallel I/O performance thus has a direct and significant impact on application performance. Because supercomputer I/O systems are large and complex, one cannot directly analyze their activity traces. While several visual or automated analysis tools for large-scale HPC log data exist, analysis research in the high-performance computing field is geared toward computation performance rather than I/O performance. Additionally, existing methods usually do not capture the network characteristics of HPC I/O systems. We present a visual analysis method for I/O trace data that takes into account the fact that HPC I/O systems can be represented as networks. We illustrate performance metrics in a way that facilitates the identification of abnormal behavior or performance problems. We demonstrate our approach on I/O traces collected from existing systems at different scales.

High quality global illumination can enhance the visual perception of depth cue and local thickness of volumetric data but it is seldom used in scientific visualization because of its high computational cost. This paper presents a novel grid-based illumination technique which is specially designed and optimized for volume visualization purpose. It supports common light sources and dynamic transfer function editing. Our method models light propagation, including both absorption and scattering, in a volume using a convection-diffusion equation that can be solved numerically. The main advantage of such technique is that the light modeling and simulation can be separated, where we can use a unified partial-differential equation to model various illumination effects, and adopt highly-parallelized grid-based numerical schemes to solve it. Results show that our method can achieve high quality volume illumination with dynamic color and opacity mapping and various light sources in real-time. The added illumination effects can greatly enhance the visual perception of spatial structures of volume data.

Abstract Large‐scale scientific simulations require execution on parallel computing systems in order to yield useful results in a reasonable time frame. But parallel execution adds communication overhead. The impact that this overhead has on performance may be difficult to gauge, as parallel application behaviors are typically harder to understand than the sequential types. We introduce an animation‐based interactive visualization technique for the analysis of communication patterns occurring in parallel application execution. Our method has the advantages of illustrating the dynamic communication patterns in the system as well as a static image of MPI (Message Passing Interface) utilization history. We also devise a data streaming mechanism that allows for the exploration of very large data sets. We demonstrate the effectiveness of our approach scaling up to 16 thousand processes using a series of trace data sets of ScaLAPACK matrix operations functions.

Abstract In this paper, we present a new method for removing shadows from images. First, shadows are detected by interactive brushing assisted with a Gaussian Mixture Model. Secondly, the detected shadows are removed using an adaptive illumination transfer approach that accounts for the reflectance variation of the image texture. The contrast and noise levels of the result are then improved with a multi‐scale illumination transfer technique. Finally, any visible shadow boundaries in the image can be eliminated based on our Bayesian framework. We also extend our method to video data and achieve temporally consistent shadow‐free results.

Abstract Geodesic grids are commonly used to model the surface of a sphere and are widely applied in numerical simulations of geoscience applications. These applications range from biodiversity, to climate change and to ocean circulation. Direct volume rendering of scalar fields defined on a geodesic grid facilitates scientists in visually understanding their large scale data. Previous solutions requiring to first transform the geodesic grid into another grid structure (e.g., hexahedral or tetrahedral grid) for using graphics hardware are not acceptable for large data, because such approaches incur significant computing and storage overhead. In this paper, we present a new method for efficient ray casting of geodesic girds by leveraging the power of Graphics Processing Units (GPUs). A geodesic grid can be directly fetched from storage or streamed from simulations to the rendering stage without the need of any intermediate grid transformation. We have designed and implemented a new analytic scheme to efficiently perform value interpolation for ray integration and gradient calculations for lighting. This scheme offers a more cost‐effective rendering solution over the existing direct rendering approach. We demonstrate the effectiveness of our rendering solution using real‐world geoscience data.

We introduce a novel spatio-temporal extrapolation technique for fluid simulation designed to improve the results without using higher resolution simulation grids. In general, there are rigid demands associated with pushing fluid animations to higher resolutions given limited computational capabilities. This results in tradeoffs between implementing high-order numerical methods and increasing the resolution of the simulation in space and time. For 3D problems, such challenges rapidly become cost-ineffective. The extrapolation method we present improves the flow features without using higher resolution simulation grids. In this paper, we show that simulation results from our extrapolation are comparable to those from higher resolution simulations. In addition, our method differs from high-order numerical methods because it does not depend on the equation or specific solver. We demonstrate that it is easy to implement and can significantly improve the fluid animation results.

Scatterplots remain a powerful tool to visualize multidimensional data. However, accurately understanding the shape of multidimensional points from 2D projections remains challenging due to overlap. Consequently, there are a lot of variations on the scatterplot as a visual metaphor for this limitation. An important aspect often overlooked in scatterplots is the issue of sensitivity or local trend, which may help in identifying the type of relationship between two variables. However, it is not well known how or what factors influence the perception of trends from 2D scatterplots. To shed light on this aspect, we conducted an experiment where we asked people to directly draw the perceived trends on a 2D scatterplot. We found that augmenting scatterplots with local sensitivity helps to fill the gaps in visual perception while retaining the simplicity and readability of a 2D scatterplot. We call this augmentation the generalized sensitivity scatterplot (GSS). In a GSS, sensitivity coefficients are visually depicted as flow lines, which give a sense of continuity and orientation of the data that provide cues about the way data points are scattered in a higher dimensional space. We introduce a series of glyphs and operations that facilitate the analysis of multidimensional data sets using GSS, and validate with a number of well-known data sets for both regression and classification tasks.

We present an integrated camera motion design and path generation system for building volume data animations. Creating animations is an essential task in presenting complex scientific visualizations. Existing visualization systems use an established animation function based on keyframes selected by the user. This approach is limited in providing the optimal in-between views of the data. Alternatively, computer graphics and virtual reality camera motion planning is frequently focused on collision free movement in a virtual walkthrough. For semi-transparent, fuzzy, or blobby volume data the collision free objective becomes insufficient. Here, we provide a set of essential criteria focused on computing camera paths to establish effective animations of volume data. Our dynamic multi-criteria solver coupled with a force-directed routing algorithm enables rapid generation of camera paths. Once users review the resulting animation and evaluate the camera motion, they are able to determine how each criterion impacts path generation. In this paper, we demonstrate how incorporating this animation approach with an interactive volume visualization system reduces the effort in creating context-aware and coherent animations. This frees the user to focus on visualization tasks with the objective of gaining additional insight from the volume data.

Visualization resizing is useful for many applications where users may use different display devices. General resizing techniques (e.g., uniform scaling) and image-resizing techniques suffer from several drawbacks, as they do not consider the content of the visualizations. This work introduces ViSizer, a perception-based framework for automatically resizing a visualization to fit any display. We formulate an energy function based on a perception model (feature congestion), which aims to determine the optimal deformation for every local region. We subsequently transform the problem into an optimization problem by the energy function. An efficient algorithm is introduced to iteratively solve the problem, allowing for automatic visualization resizing.

We present a framework for precomputed volume radiance transfer that achieves real-time rendering of global illumination effects for volume data sets such as multiple scattering, volumetric shadows, and so on. Our approach incorporates the volumetric photon mapping method into the classical precomputed radiance transfer pipeline. We contribute several techniques for light approximation, radiance transfer precomputation, and real-time radiance estimation, which are essential to make the approach practical and to achieve high frame rates. For light approximation, we propose a new discrete spherical function that has better performance for construction and evaluation when compared with existing rotational invariant spherical functions such as spherical harmonics and spherical radial basis functions. In addition, we present a fast splatting-based radiance transfer precomputation method and an early evaluation technique for real-time radiance estimation in the clustered principal component analysis space. Our techniques are validated through comprehensive evaluations and rendering tests. We also apply our rendering approach to volume visualization.

With the evolution of graphics hardware, high quality global illumination becomes available for real-time volume rendering. Compared to local illumination, global illumination can produce realistic shading effects which are closer to real world scenes, and has proven useful for enhancing volume data visualization to enable better depth and shape perception. However, setting up optimal lighting could be a nontrivial task for average users. There were lighting design works for volume visualization but they did not consider global light transportation. In this paper, we present a lighting design method for volume visualization employing global illumination. The resulting system takes into account view and transfer-function dependent content of the volume data to automatically generate an optimized three-point lighting environment. Our method fully exploits the back light which is not used by previous volume visualization systems. By also including global shadow and multiple scattering, our lighting system can effectively enhance the depth and shape perception of volumetric features of interest. In addition, we propose an automatic tone mapping operator which recovers visual details from overexposed areas while maintaining sufficient contrast in the dark areas. We show that our method is effective for visualizing volume datasets with complex structures. The structural information is more clearly and correctly presented under the automatically generated light sources.

Visualizing complex volume data usually renders selected parts of the volume semitransparently to see inner structures of the volume or provide a context. This presents a challenge for volume rendering methods to produce images with unambiguous depth-ordering perception. Existing methods use visual cues such as halos and shadows to enhance depth perception. Along with other limitations, these methods introduce redundant information and require additional overhead. This paper presents a new approach to enhancing depth-ordering perception of volume rendered images without using additional visual cues. We set up an energy function based on quantitative perception models to measure the quality of the images in terms of the effectiveness of depth-ordering and transparency perception as well as the faithfulness of the information revealed. Guided by the function, we use a conjugate gradient method to iteratively and judiciously enhance the results. Our method can complement existing systems for enhancing volume rendering results. The experimental results demonstrate the usefulness and effectiveness of our approach.

After the news of Osama Bin Laden's death leaked through Twitter, many people wondered if Twitter would fundamentally change the way we produce, spread, and consume news. In this paper we provide an in-depth analysis of how the news broke and spread on Twitter. We confirm the claim that Twitter broke the news first, and find evidence that Twitter had convinced a large number of its audience before mainstream media confirmed the news. We also discover that attention on Twitter was highly concentrated on a small number of "opinion leaders" and identify three groups of opinion leaders who played key roles in spreading the news: individuals affiliated with media played a large part in breaking the news, mass media brought the news to a wider audience and provided eager Twitter users with content on external sites, and celebrities helped to spread the news and stimulate conversation. Our findings suggest Twitter has great potential as a news medium.

Modern large-scale heterogeneous computers incorporating GPUs offer impressive processing capabilities. It is desirable to fully utilize such systems for serving multiple users concurrently to visualize large data at interactive rates. However, as the disparity between data transfer speed and compute speed continues to increase in heterogeneous systems, data locality becomes crucial for performance. We present a new job scheduling design to support multi-user exploration of large data in a heterogeneous computing environment, achieving near optimal data locality and minimizing I/O overhead. The targeted application is a parallel visualization system which allows multiple users to render large volumetric data sets in both interactive mode and batch mode. We present a cost model to assess the performance of parallel volume rendering and quantify the efficiency of job scheduling. We have tested our job scheduling scheme on two heterogeneous systems with different configurations. The largest test volume data used in our study has over two billion grid points. The timing results demonstrate that our design effectively improves data locality for complex multi-user job scheduling problems, leading to better overall performance of the service.

Sparse linear support vector machines have been widely applied to variable selection in many applications. For large data, managing the cost of training a sparse model with good predication performance is an essential topic. In this work, we propose a scalable training algorithm for large-scale data with millions of examples and features. We develop a dual alternating direction method for solving L1-regularized linear SVMs. The learning procedure simply involves quadratic programming in the same form as the standard SVM dual, followed by a soft-thresholding operation. The proposed training algorithm possesses two favorable properties. First, it is a decomposable algorithm by which a large problem can be reduced to small ones. Second, the sparsity of intermediate solutions is maintained throughout the training process. It naturally promotes the solution sparsity by soft-thresholding. We demonstrate that, by experiments, our method outperforms state-of-the-art approaches on large-scale benchmark data sets. We also show that it is well suited for training large sparse models on a distributed system.

Anonymized Call Detail Records (CDRs) contain positional information of large populations and therefore have been extensively analyzed to understand human mobility. Due to the temporally sparse and spatially coarse nature of the data, most of these studies have focused on primitive aspects of movements such as travel distance and speed. Incorporating underlying geographic information in these analyses would allow analysts to put these movements into context and to gain deeper insight into how metropolitan areas function. In this paper, we present a set of procedures for inferring mobile users' mobility patterns while retaining the context of underlying geography. We apply these methods to our case study on New York City anonymized CDRs. We find that our methods verify current areal semantics and commuting rush-hour patterns, and we also derive further implications regarding geographic, demographic, and other effects on human mobility.

In this poster, we present a volume rendering framework that achieves realtime rendering of global illumination effects for volume datasets, such as multiple scattering and volume shadow. This approach incorporates the volumetric photon mapping technique [Jensen and Christensen 1998] into the classical precomputed radiance transfer [Sloan et al. 2002] pipeline. Fig.1 shows that our method is successfully applied in both interactive graphics and scientific visualization applications.

In this paper, we study the sensitivity of centrality metrics as a key metric of social networks to support visual reasoning. As centrality represents the prestige or importance of a node in a network, its sensitivity represents the importance of the relationship between this and all other nodes in the network. We have derived an analytical solution that extracts the sensitivity as the derivative of centrality with respect to degree for two centrality metrics based on feedback and random walks. We show that these sensitivities are good indicators of the distribution of centrality in the network, and how changes are expected to be propagated if we introduce changes to the network. These metrics also help us simplify a complex network in a way that retains the main structural properties and that results in trustworthy, readable diagrams. Sensitivity is also a key concept for uncertainty analysis of social networks, and we show how our approach may help analysts gain insight on the robustness of key network metrics. Through a number of examples, we illustrate the need for measuring sensitivity, and the impact it has on the visualization of and interaction with social and other scale-free networks.

In this work, we address the problem of lossless compression of scientific and medical floating-point volume data. We propose two prediction-based compression methods that share a common framework, which consists of a switched prediction scheme wherein the best predictor out of a preset group of linear predictors is selected. Such a scheme is able to adapt to different datasets as well as to varying statistics within the data. The first method, called APE (Adaptive Polynomial Encoder), uses a family of structured interpolating polynomials for prediction, while the second method, which we refer to as ACE (Adaptive Combined Encoder), combines predictors from previous work with the polynomial predictors to yield a more flexible, powerful encoder that is able to effectively decorrelate a wide range of data. In addition, in order to facilitate efficient visualization of compressed data, our scheme provides an option to partition floating-point values in such a way as to provide a progressive representation. We compare our two compressors to existing state-of-the-art lossless floating-point compressors for scientific data, with our data suite including both computer simulations and observational measurements. The results demonstrate that our polynomial predictor, APE, is comparable to previous approaches in terms of speed but achieves better compression rates on average. ACE, our combined predictor, while somewhat slower, is able to achieve the best compression rate on all datasets, with significantly better rates on most of the datasets.

In order to assess the reliability of volume rendering, it is necessary to consider the uncertainty associated with the volume data and how it is propagated through the volume rendering algorithm, as well as the contribution to uncertainty from the rendering algorithm itself. In this work, we show how to apply concepts from the field of reliable computing in order to build a framework for management of uncertainty in volume rendering, with the result being a self-validating computational model to compute a posteriori uncertainty bounds. We begin by adopting a coherent, unifying possibility-based representation of uncertainty that is able to capture the various forms of uncertainty that appear in visualization, including variability, imprecision, and fuzziness. Next, we extend the concept of the fuzzy transform in order to derive rules for accumulation and propagation of uncertainty. This representation and propagation of uncertainty together constitute an automated framework for management of uncertainty in visualization, which we then apply to volume rendering. The result, which we call fuzzy volume rendering, is an uncertainty-aware rendering algorithm able to produce more complete depictions of the volume data, thereby allowing more reliable conclusions and informed decisions. Finally, we compare approaches for self-validated computation in volume rendering, demonstrating that our chosen method has the ability to handle complex uncertainty while maintaining efficiency.

We present a two-level approach for height map estimation from single images, aiming at restoring brick and stone relief(BSR) from their rubbing images in a visually plausible manner. In our approach, the base relief of the low frequency component is estimated automatically with a partial differential equation (PDE)-based mesh deformation scheme. A few vertices near the central area of the object region are selected and assigned with heights estimated by an erosion-based contour map. These vertices together with object boundary vertices, boundary normals as well as the partial differential properties of the mesh are taken as constraints to deform the mesh by minimizing a least-squares error functional. The high frequency detail is estimated directly from rubbing images automatically or optionally with minimal interactive processing. The final height map for a restored BSR is obtained by blending height maps of the base relief and high frequency detail. We demonstrate that our method can not only successfully restore several BSR maps from their rubbing images, but also restore some relief-like surfaces from photographic images.

Graph visualization has been widely used to understand and present both global structural and local adjacency information in relational data sets (e.g., transportation networks, citation networks, or social networks). Graphs with dense edges, however, are difficult to visualize because fast layout and good clarity are not always easily achieved. When the number of edges is large, edge bundling can be used to improve the clarity, but in many cases, the edges could be still too cluttered to permit correct interpretation of the relations between nodes. In this paper, we present an ambiguity-free edge-bundling method especially for improving local detailed view of a complex graph. Our method makes more efficient use of display space and supports detail-on-demand viewing through an interactive interface. We demonstrate the effectiveness of our method with public coauthorship network data.

Interactive visualizations can allow science museum visitors to explore new worlds by seeing and interacting with scientific data. However, designing interactive visualizations for informal learning environments, such as museums, presents several challenges. First, visualizations must engage visitors on a personal level. Second, visitors often lack the background to interpret visualizations of scientific data. Third, visitors have very limited time at individual exhibits in museums. This paper examines these design considerations through the iterative development and evaluation of an interactive exhibit as a visualization tool that gives museumgoers access to scientific data generated and used by researchers. The exhibit prototype, Living Liquid, encourages visitors to ask and answer their own questions while exploring the time-varying global distribution of simulated marine microbes using a touchscreen interface. Iterative development proceeded through three rounds of formative evaluations using think-aloud protocols and interviews, each round informing a key visualization design decision: (1) what to visualize to initiate inquiry, (2) how to link data at the microscopic scale to global patterns, and (3) how to include additional data that allows visitors to pursue their own questions. Data from visitor evaluations suggests that, when designing visualizations for public audiences, one should (1) avoid distracting visitors from data that they should explore, (2) incorporate background information into the visualization, (3) favor understandability over scientific accuracy, and (4) layer data accessibility to structure inquiry. Lessons learned from this case study add to our growing understanding of how to use visualizations to actively engage learners with scientific data.

Storyline visualization is a technique used to depict the temporal dynamics of social interactions. This visualization technique was first introduced as a hand-drawn illustration in XKCD's "Movie Narrative Charts" [21]. If properly constructed, the visualization can convey both global trends and local interactions in the data. However, previous methods for automating storyline visualizations are overly simple, failing to achieve some of the essential principles practiced by professional illustrators. This paper presents a set of design considerations for generating aesthetically pleasing and legible storyline visualizations. Our layout algorithm is based on evolutionary computation, allowing us to effectively incorporate multiple objective functions. We show that the resulting visualizations have significantly improved aesthetics and legibility compared to existing techniques.

Uncertainty can arise in any stage of a visual analytics process, especially in data-intensive applications with a sequence of data transformations. Additionally, throughout the process of multidimensional, multivariate data analysis, uncertainty due to data transformation and integration may split, merge, increase, or decrease. This dynamic characteristic along with other features of uncertainty pose a great challenge to effective uncertainty-aware visualization. This paper presents a new framework for modeling uncertainty and characterizing the evolution of the uncertainty information through analytical processes. Based on the framework, we have designed a visual metaphor called uncertainty flow to visually and intuitively summarize how uncertainty information propagates over the whole analysis pipeline. Our system allows analysts to interact with and analyze the uncertainty information at different levels of detail. Three experiments were conducted to demonstrate the effectiveness and intuitiveness of our design.

Finding correlations among data is one of the most essential tasks in many scientific investigations and discoveries. This paper addresses the issue of creating a static volume classification that summarizes the correlation connection in time-varying multivariate data sets. In practice, computing all temporal and spatial correlations for large 3D time-varying multivariate data sets is prohibitively expensive. We present a sampling-based approach to classifying correlation patterns. Our sampling scheme consists of three steps: selecting important samples from the volume, prioritizing distance computation for sample pairs, and approximating volume-based correlation with sample-based correlation. We classify sample voxels to produce static visualization that succinctly summarize the connection among all correlation volumes with respect to various reference locations. We also investigate the error introduced by each step of our sampling scheme in terms of classification accuracy. Domain scientists participated in this work and helped us select samples and evaluate results. Our approach is generally applicable to the analysis of other scientific data where correlation study is relevant.

Advanced computing and imaging technologies enable scientists to study natural and physical phenomena at unprecedented precision, resulting in an explosive growth of data. Furthermore, the size of the collected information about the Internet and mobile device users is expected to be even greater, a daunting challenge we must address in order to make sense and maximize utilization of all the available information for decision making and knowledge discovery. I will introduce a few new approaches to large data visualization for revealing hidden structures and gleaning insights from large, complex data found in many areas of study.

Effective analysis and visualization of time-varying multivariate data is crucial for understanding complex and dynamic variable interaction and temporal evolution. Advances made in this area are mainly on query-driven visualization and correlation exploration. Solutions and techniques that investigate the important aspect of causal relationships among variables have not been sought. In this paper, we present a new approach to analyzing and visualizing time-varying multivariate volumetric and particle data sets through the study of information flow using the information-theoretic concept of transfer entropy. We employ time plot and circular graph to show information transfer for an overview of relations among all pairs of variables. To intuitively illustrate the influence relation between a pair of variables in the visualization, we modulate the color saturation and opacity for volumetric data sets and present three different visual representations, namely, ellipse, smoke, and metaball, for particle data sets. We demonstrate this information-theoretic approach and present our findings with three time-varying multivariate data sets produced from scientific simulations.

Current simulations of turbulent flames are instrumented with particles to capture the dynamic behavior of combustion in next-generation engines. Categorizing the set of many millions of particles, each of which is featured with a history of its movement positions and changing thermo-chemical states, helps understand the turbulence mechanism. We introduce a dual-space method to analyze such data, starting by clustering the time series curves in the phase space of the data, and then visualizing the corresponding trajectories of each cluster in the physical space. To cluster time series curves, we adopt a model-based clustering technique in a two-stage scheme. In the first stage, the characteristics of shape and relative position are particularly concerned in classifying the time series curves, and in the second stage, within each group of curves, clustering is further conducted based on how the curves change over time. In our work, we perform the model-based clustering in a semi-supervised manner. Users' domain knowledge is integrated through intuitive interaction tools to steer the clustering process. Our dual-space method has been used to analyze particle data in combustion simulations and can also be applied to other scientific simulations involving particle trajectory analysis work.

Simulating the expansion of a Type II supernova using an adaptive computational fluid dynamics (CFD) engine yields a complex mixture of turbulent flow with dozens of physical properties. The dataset shown in this sketch was initially simulated on iVEC's EPIC supercomputer (a 9600 core Linux cluster) using FLASH [Fryxell et al. 2000] to model the thermonuclear explosion, and later post-processed using a novel integration technique to derive the radio frequency emission spectra of the expanding shock-wave front [Potter et al. 2011]. Model parameters have been chosen to simulate the asymmetric properties of the SN 1987A remnant [Potter et al. 2009].

Monitoring, anomaly detection and forensics are essential tasks that must be carried out routinely for every computer network. The sheer volume of data generated by conventional anomaly detection tools such as Snort often makes it difficult to explain the nature of an attack and track down its source. In this paper we present TVi, a tool that combines multiple visual representations of network traces carefully designed and tightly coupled to support different levels of visual-based querying and reasoning required for making sense of complex traffic data. TVi allows analysts to visualize data starting at a high level, providing information related to the entire network, and easily move all the way down to a very low level, providing detailed information about selected hosts, anomalies and attack paths. We designed TVi with scalability and extensibility in mind: its DBMS foundations make it scalable with virtually no limitations, and other state-of-the-art IDS, like Snort or Bro, can be easily integrated in our tool. We demonstrate with two case studies, a synthetic dataset (DARPA 1999) and a real one (University of Brescia, UniBS, 2009), how TVi can enhance a network administrator's ability to reveal hidden patterns in network traces and link their key information so as to easily reveal details that by merely observing Snort's output would go unnoticed. We make TVi's source code available to the community under an Open Source license.

Abstract Most 3D vector field visualization techniques suffer from the problem of visual clutter, and it remains a challenging task to effectively convey both directional and structural information of 3D vector fields. In this paper, we present a novel visualization framework that combines the advantages of clustering methods and illustrative rendering techniques to generate a concise and informative depiction of complex flow structures. Given a 3D vector field, we first generate a number of streamlines covering the important regions based on an entropy measurement. Then we decompose the streamlines into different groups based on a categorization of vector information, wherein the streamline pattern in each group is ensured to be coherent or nearly coherent. For each group, we select a set of representative streamlines and render them in an illustrative fashion to enhance depth cues and succinctly show local flow characteristics. The results demonstrate that our approach can generate a visualization that is relatively free of visual clutter while facilitating perception of salient information of complex vector fields.

Abstract Understanding large, complex networks is important for many critical tasks, including decision making, process optimization, and threat detection. Existing network analysis tools often lack intuitive interfaces to support the exploration of large scale data. We present a visual recommendation system to help guide users during navigation of network data. Collaborative filtering, similarity metrics, and relative importance are used to generate recommendations of potentially significant nodes for users to explore. In addition, graph layout and node visibility are adjusted in real‐time to accommodate recommendation display and to reduce visual clutter. Case studies are presented to show how our design can improve network exploration.

Abstract Word clouds are proliferating on the Internet and have received much attention in visual analytics. Although word clouds can help users understand the major content of a document collection quickly, their ability to visually compare documents is limited. This paper introduces a new method to create semantic‐preserving word clouds by leveraging tailored seam carving, a well‐established content‐aware image resizing operator. The method can optimize a word cloud layout by removing a left‐to‐right or top‐to‐bottom seam iteratively and gracefully from the layout. Each seam is a connected path of low energy regions determined by a Gaussian‐based energy function. With seam carving, we can pack the word cloud compactly and effectively, while preserving its overall semantic structure. Furthermore, we design a set of interactive visualization techniques for the created word clouds to facilitate visual text analysis and comparison. Case studies are conducted to demonstrate the effectiveness and usefulness of our techniques.

Images that seamlessly combine views at different levels of detail are appealing. However, creating such multiscale images is not a trivial task, and most such illustrations are handcrafted by skilled artists. This paper presents a framework for direct multiscale rendering of geometric and volumetric models. The basis of our approach is a set of non-linearly bent camera rays that smoothly cast through multiple scales. We show that by properly setting up a sequence of conventional pinhole cameras to capture features of interest at different scales, along with image masks specifying the regions of interest for each scale on the projection plane, our rendering framework can generate non-linear sampling rays that smoothly project objects in a scene at multiple levels of detail onto a single image. We address two important issues with non-linear camera projection. First, our streamline-based ray generation algorithm avoids undesired camera ray intersections, which often result in unexpected images. Second, in order to maintain camera ray coherence and preserve aesthetic quality, we create an interpolated 3D field that defines the contribution of each pinhole camera for determining ray orientations. The resulting multiscale camera has three main applications: (1) presenting hierarchical structure in a compact and continuous manner, (2) achieving focus+context visualization, and (3) creating fascinating and artistic images.

This paper presents a study of gradient estimation methods for rendering unstructured-mesh volume data. Gradient estimation is necessary for rendering shaded isosurfaces and specular highlights, which provide important cues for shape and depth. Gradient estimation has been widely studied and deployed for regular-grid volume data to achieve local illumination effects, but has been, otherwise, for unstructured-mesh data. As a result, most of the unstructured-mesh volume visualizations made so far were unlit. In this paper, we present a comprehensive study of gradient estimation methods for unstructured meshes with respect to their cost and performance. Through a number of benchmarks, we discuss the effects of mesh quality and scalar function complexity in the accuracy of the reconstruction, and their impact in lighting-enabled volume rendering. Based on our study, we also propose two heuristic improvements to the gradient reconstruction process. The first heuristic improves the rendering quality with a hybrid algorithm that combines the results of the multiple reconstruction methods, based on the properties of a given mesh. The second heuristic improves the efficiency of its GPU implementation, by restricting the computation of the gradient on a fixed-size local neighborhood.

Direct volume rendering is an important tool for visualizing complex data sets. However, in the process of generating 2D images from 3D data, information is lost in the form of attenuation and occlusion. The lack of a feedback mechanism to quantify the loss of information in the rendering process makes the design of good transfer functions a difficult and time consuming task. In this paper, we present the general notion of visibility histograms, which are multidimensional graphical representations of the distribution of visibility in a volume-rendered image. In this paper, we explore the 1D and 2D transfer functions that result from intensity values and gradient magnitude. With the help of these histograms, users can manage a complex set of transfer function parameters that maximize the visibility of the intervals of interest and provide high quality images of volume data. We present a semiautomated method for generating transfer functions, which progressively explores the transfer function space toward the goal of maximizing visibility of important structures. Our methodology can be easily deployed in most visualization systems and can be used together with traditional 1D and 2D opacity transfer functions based on scalar values, as well as with other more sophisticated rendering algorithms.

The growing sizes of volumetric data sets pose a great challenge for interactive visualization. In this paper, we present a feature-preserving data reduction and focus+context visualization method based on transfer function driven, continuous voxel repositioning and resampling techniques. Rendering reduced data can enhance interactivity. Focus+context visualization can show details of selected features in context on display devices with limited resolution. Our method utilizes the input transfer function to assign importance values to regularly partitioned regions of the volume data. According to user interaction, it can then magnify regions corresponding to the features of interest while compressing the rest by deforming the 3D mesh. The level of data reduction achieved is significant enough to improve overall efficiency. By using continuous deformation, our method avoids the need to smooth the transition between low and high-resolution regions as often required by multiresolution methods. Furthermore, it is particularly attractive for focus+context visualization of multiple features. We demonstrate the effectiveness and efficiency of our method with several volume data sets from medical applications and scientific simulations.

This paper presents visualization of field-measured, time-varying multidimensional earthquake accelerograph readings. Direct volume rendering is used to depict the space-time relationships of seismic readings collected from sensor stations in an intuitive way such that the progress of seismic wave propagation of an earthquake event can be directly observed. The resulting visualization reveals the sequence of seismic wave initiation, propagation, attenuation over time, and energy releasing events. We provide a case study on the magnitude scale Mw7.6 Chi-Chi earthquake in Taiwan, which is the most thoroughly recorded earthquake event ever in the history. More than 400 stations recorded this event, and the readings from this event increased global strong-motion records five folds. Each station measured east-west, north-south, and vertical component of acceleration for approximately 90 seconds. The sensor network released the initial raw data within minutes after the Chi-Chi mainshock. It is essential to have a visualization system for fast data exploring and analyzing, offering crucial visual analytical information for scientists to make quick judgments. Raw data requires preprocessing before it can be rendered. We generated a sequence of ground-motion wave-field maps of 350 × 200 regular grid covers the entire Taiwan island from the sensor network readings. The result is a total of 1000 ground-motion wave-field maps with 0.1 second interval, forming a 1000 × 350 × 200 volume data set. We show that visualizing the time-varying component of the data spatially uncovers the changing features hidden in the data.

We present a technique which automatically converts a small number of single-view volume rendered images of the same 3D data set into a compact representation of that data set. This representation is a multi-layered image, or an explorable image, which enables interactive exploration of volume data in transfer function space without accessing the original data. We achieve this by automatically extracting layers depicted in composited images. The layers can then be recombined in different ways to simulate opacity changes and recoloring of individual features. Our results demonstrate that explorable images are especially useful when the volume data is too large for interactive exploration, takes too long to render due to the underlying mesh structure or desired shading effect, or if the original volume data is not available. Explorable images can offer real-time image-based interaction as a preview mechanism for remote visualization or visualization of large volume data on low-end hardware, within a mobile device, or a Web browser.

In flow visualization, field lines are often used to convey both global and local structure and movement of the flow. One challenge is to find and classify the representative field lines. Most existing solutions follow an automatic approach that generates field lines characterizing the flow and arranges these lines into a single picture. In our work, we advocate a user-centric approach to exploring 3D vector fields. Our method allows the user to sketch 2D curves for pattern matching in 2D and field lines clustering in 3D. Specifically, a 3D field line whose view-dependent 2D projection is most similar to the user drawing will be identified and utilized to extract all similar 3D field lines. Furthermore, we employ an automatic clustering method to generate field-line templates for the user to locate subfields of interest. This semi-automatic process leverages the user's knowledge about the flow field through intuitive user interaction, resulting in a promising alternative to existing flow visualization solutions. With our sketch-based interface, the user can effectively dissect the flow field and make more structured visualization for analysis or presentation.

The pervasive concept of cloud computing suggests that visualization, which is both data and computing intensive, is a perfect cloud computing application. This paper presents a sketch of an interface design for an online visualization service. To make such a service attractive to a wider audience, its user interface must be simple and easy to use for both casual and expert users. We envision an interface that supports visualization processes mainly directed by browsing and assessing existing visualizations in terms of images and videos will be very appealing to, in particular, casual users. That is, the aim is to maximize the utilization of the rich visualization data on the web. Without losing generality, we consider volume data visualization applications for our interface design. We also discuss issues in organizing online visualization data, and constructing and managing a rendering cloud.

In this paper we present a multi-GPU parallel volume rendering implemention built using the MapReduce programming model. We give implementation details of the library, including specific optimizations made for our rendering and compositing design. We analyze the theoretical peak performance and bottlenecks for all tasks required and show that our system significantly reduces computation as a bottleneck in the ray-casting phase. We demonstrate that our rendering speeds are adequate for interactive visualization (our system is capable of rendering a 10243 floating-point sampled volume in under one second using 8 GPUs), and that our system is capable of delivering both in-core and out-of-core visualizations. We argue that a multi-GPU MapReduce library is a good fit for parallel volume renderering because it is easy to program for, scales well, and eliminates the need to focus on I/O algorithms thus allowing the focus to be on visualization algorithms instead. We show that our system scales with respect to the size of the volume, and (given enough work) the number of GPUs.

As a user's digital music collection grows, it can become difficult to navigate. Music library programs aid in this task by organizing music according to tags such as artist or title. However these generally utilize a text based interface, and they do not take into account the content of the music itself. As such, they do not handle untagged or mistagged music well. Automated metrics exist, but are not as widely used since they have the potential to be unreliable. This paper presents a graph-based visual interface for exploring a library of music based on analysis of the content of the music rather than tag information, which allows the user to navigate a music library thematically.

Abstract The selection of an appropriate global transfer function is essential for visualizing time‐varying simulation data. This is especially challenging when the global data range is not known in advance, as is often the case in remote and in‐situ visualization settings. Since the data range may vary dramatically as the simulation progresses, volume rendering using local transfer functions may not be coherent for all time steps. We present an exploratory technique that enables coherent classification of time‐varying volume data. Unlike previous approaches, which require pre‐processing of all time steps, our approach lets the user explore the transfer function space without accessing the original 3D data. This is useful for interactive visualization, and absolutely essential for in‐situ visualization, where the entire simulation data range is not known in advance. Our approach generates a compact representation of each time step at rendering time in the form of ray attenuation functions, which are used for subsequent operations on the opacity and color mappings. The presented approach offers interactive exploration of time‐varying simulation data that alleviates the cost associated with reloading and caching large data sets.

Abstract Strange attractors of 3D vector field flows sometimes have a fractal geometric structure in one dimension, and smooth surface behavior in the other two. General flow visualization methods show the flow dynamics well, but not the fractal structure. Here we approximate the attractor by polygonal surfaces, which reveal the fractal geometry. We start with a polygonal approximation which neglects the fractal dimension, and then deform it by the flow to create multiple sheets of the fractal structure. We use adaptive subdivision, mesh decimation, and retiling methods to preserve the quality of the polygonal surface in the face of extreme stretching, bending, and creasing caused by the flow. A GPU implementation provides efficient visualization, which we also apply to other turbulent flows.

This paper presents a system for generating dynamic narratives from videos. These narratives are characterized for being compact, coherent and interactive, as inspired by principles of sequential art. Narratives depict the motion of one or several actors over time. Creating compact narratives is challenging as it is desired to combine the video frames in a way that reuses redundant backgrounds and depicts the stages of a motion. In addition, previous approaches focus on the generation of static summaries and can afford expensive image composition techniques. A dynamic narrative, on the other hand, must be played and skimmed in real-time, which imposes certain cost limitations in the video processing. In this paper, we define a novel process to compose foreground and background regions of video frames in a single interactive image using a series of spatio-temporal masks. These masks are created to improve the output of automatic video processing techniques such as image stitching and foreground segmentation. Unlike hand-drawn narratives, often limited to static representations, the proposed system allows users to explore the narrative dynamically and produce different representations of motion. We have built an authoring system that incorporates these methods and demonstrated successful results on a number of video clips. The authoring system can be used to create interactive posters of video clips, browse video in a compact manner or highlight a motion sequence in a movie.

In flow simulations the behavior and properties of particle trajectories often depend on the physical geometry contained in the simulated environment. Understanding the flow in and around the geometry itself is an important part of analyzing the data. Previous work has often utilized focus+context rendering techniques, with an emphasis on showing trajectories while simplifying or illustratively rendering the physical areas. Our research instead emphasizes the local relationship between particle paths and geometry by using a projected multi-field visualization technique. The correlation between a particle path and its surrounding area is calculated on-the-fly and displayed in a non-intrusive manner. In addition, we support visual exploration and comparative analysis through the use of linked information visualization, such as manipulatable curve plots and one-on-one similarity plots. Our technique is demonstrated on particle trajectories from a groundwater simulation and a computer room airflow simulation, where the flow of particles is highly influenced by the dense geometry.

This paper introduces a new streamline placement and selection algorithm for 3D vector fields. Instead of considering the problem as a simple feature search in data space, we base our work on the observation that most streamline fields generate a lot of self-occlusion which prevents proper visualization. In order to avoid this issue, we approach the problem in a view-dependent fashion and dynamically determine a set of streamlines which contributes to data understanding without cluttering the view. Since our technique couples flow characteristic criteria and view-dependent streamline selection we are able achieve the best of both worlds: relevant flow description and intelligible, uncluttered pictures. We detail an efficient GPU implementation of our algorithm, show comprehensive visual results on multiple datasets and compare our method with existing flow depiction techniques. Our results show that our technique greatly improves the readability of streamline visualizations on different datasets without requiring user intervention.

Interactivity is key to exploration of volume data. Interactivity may be hindered due to many factors, e.g. large data size,high resolution or complexity of a data set, or an expensive rendering algorithm. We present a novel framework for visualizing volume data that enables interactive exploration using proxy images, without accessing the original 3D data. Data exploration using direct volume rendering requires multiple (often redundant) accesses to possibly large amounts of data. The notion of visualization by proxy relies on the ability to defer operations traditionally used for exploring 3D data to a more suitable intermediate representation for interaction--proxy images. Such operations include view changes, transfer function exploration, and relighting. While previous work has addressed specific interaction needs, we provide a complete solution that enables real-time interaction with large data sets and has low hardware and storage requirements.

Direct volume rendering is an important tool for visualizing complex data sets. However, in the process of generating 2D images from 3D data, information is lost in the form of attenuation and occlusion. The lack of a feedback mechanism to quantify the loss of information in the rendering process makes the design of good transfer functions a difficult and time consuming task. In this paper, we present the notion of visibility-driven transfer functions, which are transfer functions that provide a good visibility of features of interest from a given viewpoint. To achieve this, we introduce visibility histograms. These histograms provide graphical cues that intuitively inform the user about the contribution of particular scalar values to the final image. By carefully manipulating the parameters of the opacity transfer function, users can now maximize the visibility of the intervals of interest in a volume data set. Based on this observation, we also propose a semi-automated method for generating transfer functions, which progressively improves a transfer function defined by the user, according to a certain importance metric. Now the user does not have to deal with the tedious task of making small changes to the transfer function parameters, but now he/she can rely on the system to perform these searches automatically. Our methodology can be easily deployed in most visualization systems and can be used together with traditional 1D opacity transfer functions based on scalar values, as well as with multidimensional transfer functions and other more sophisticated rendering algorithms.

Many graphs used in real-world applications consist of nodes belonging to more than one category. We call such graph ldquomultiple-category graphsrdquo. Social networks are typical examples of multiple-category graphs: nodes are persons, links are friendships, and categories are communities that the persons belong to. It is often helpful to visualize both connectivity and categories of the graphs simultaneously. In this paper, we present a new visualization technique for multiple-category graphs. The technique firstly constructs hierarchical clusters of the nodes based on both connectivity and categories. It then places the nodes by a new hybrid space-filling and force-directed layout algorithm to clearly display both connectivity and category information. We show layout results using our hybrid method and compare it with other methods, and present a case study using an active biological network dataset.

The ability to extract and follow time-varying flow features in volume data generated from large-scale numerical simulations enables scientists to effectively see and validate modeled phenomena and processes. Extracted features often take much less storage space and computing resources to visualize. Most feature extraction and tracking methods first identify features of interest in each time step independently, then correspond these features in consecutive time steps of the data. Since these methods handle each time step separately, they do not use the coherency of the feature along the time dimension in the extraction process. In this paper, we present a prediction-correction method that uses a prediction step to make the best guess of the feature region in the subsequent time step, followed by growing and shrinking the border of the predicted region to coherently extract the actual feature of interest. This method makes use of the temporal-space coherency of the data to accelerate the extraction process while implicitly solving the tedious correspondence problem that previous methods focus on. Our method is low cost with very little storage overhead, and thus facilitates interactive or runtime extraction and visualization, unlike previous methods which were largely suited for batch-mode processing due to high computational cost.

We present a correlation study of time-varying multivariate volumetric data sets. In most scientific disciplines, to test hypotheses and discover insights, scientists are interested in looking for connections among different variables, or among different spatial locations within a data field. In response, we propose a suite of techniques to analyze the correlations in time-varying multivariate data. Various temporal curves are utilized to organize the data and capture the temporal behaviors. To reveal patterns and find connections, we perform data clustering and segmentation using the k-means clustering and graph partitioning algorithms. We study the correlation structure of a single or a pair of variables using pointwise correlation coefficients and canonical correlation analysis. We demonstrate our approach using results on time-varying multivariate climate data sets.

This paper presents a novel methodology for social network discovery based on the sensitivity coefficients of importance metrics, namely the Markov centrality of a node, a metric based on random walks. Analogous to node importance, which ranks the important nodes in a social network, the sensitivity analysis of this metric provides a ranking of the relationships between nodes. The sensitivity parameter of the importance of a node with respect to another measures the direct or indirect impact of a node. We show that these relationships help discover hidden links between nodes and highlight meaningful links between seemingly disparate sub-networks in a social structure. We introduce the notion of implicit links, which represent an indirect relationship between nodes not connected by edges, seen as hidden connections in complex networks. We demonstrate our methodology on two social network data sets and use sensitivity-guided visualizations to highlight our findings. Our results show that this analytic tool, when coupled with visualization, is an effective mechanism for discovering social networks.

In addition to their role as simulation engines, modern supercomputers can be harnessed for scientific visualization. Their extensive concurrency, parallel storage systems, and high-performance interconnects can mitigate the expanding size and complexity of scientific datasets and prepare for in situ visualization of these data. In ongoing research into testing parallel volume rendering on the IBM Blue Gene/P (BG/P), we measure performance of disk I/O, rendering, and compositing on large datasets, and evaluate bottlenecks with respect to system-specific I/O and communication patterns. To extend the scalability of the direct-send image compositing stage of the volume rendering algorithm, we limit the number of compositing cores when many small messages are exchanged. To improve the data-loading stage of the volume renderer, we study the I/O signatures of the algorithm in detail. The results of this research affirm that a distributed-memory computing architecture such as BG/P is a scalable platform for large visualization problems.

Despite the ever-growing improvements on graphics processing units and computational power, classifying 3D volume data remains a challenge.In this paper, we present a new method for classifying volume data based on the ambient occlusion of voxels. This information stems from the observation that most volumes of a certain type, e.g., CT, MRI or flow simulation, contain occlusion patterns that reveal the spatial structure of their materials or features. Furthermore, these patterns appear to emerge consistently for different data sets of the same type. We call this collection of patterns the occlusion spectrum of a dataset. We show that using this occlusion spectrum leads to better two-dimensional transfer functions that can help classify complex data sets in terms of the spatial relationships among features. In general, the ambient occlusion of a voxel can be interpreted as a weighted average of the intensities in a spherical neighborhood around the voxel. Different weighting schemes determine the ability to separate structures of interest in the occlusion spectrum. We present a general methodology for finding such a weighting. We show results of our approach in 3D imaging for different applications, including brain and breast tumor detection and the visualization of turbulent flow.

We present two visualization techniques for curve-centric volume reformation with the aim to create compelling comparative visualizations. A curve-centric volume reformation deforms a volume, with regards to a curve in space, to create a new space in which the curve evaluates to zero in two dimensions and spans its arc-length in the third. The volume surrounding the curve is deformed such that spatial neighborhood to the curve is preserved. The result of the curve-centric reformation produces images where one axis is aligned to arc-length, and thus allows researchers and practitioners to apply their arc-length parameterized data visualizations in parallel for comparison. Furthermore we show that when visualizing dense data, our technique provides an inside out projection, from the curve and out into the volume, which allows for inspection what is around the curve. Finally we demonstrate the usefulness of our techniques in the context of two application cases. We show that existing data visualizations of arc-length parameterized data can be enhanced by using our techniques, in addition to creating a new view and perspective on volumetric data around curves. Additionally we show how volumetric data can be brought into plotting environments that allow precise readouts. In the first case we inspect streamlines in a flow field around a car, and in the second we inspect seismic volumes and well logs from drilling.

In serial computation, program profiling is often helpful for optimization of key sections of code. When moving to parallel computation, not only does the code execution need to be considered but also communication between the different processes which can induce delays that are detrimental to performance. As the number of processes increases, so does the impact of the communication delays on performance. For large-scale parallel applications, it is critical to understand how the communication impacts performance in order to make the code more efficient. There are several tools available for visualizing program execution and communications on parallel systems. These tools generally provide either views which statistically summarize the entire program execution or process-centric views. However, process-centric visualizations do not scale well as the number of processes gets very large. In particular, the most common representation of parallel processes is a Gantt char t with a row for each process. As the number of processes increases, these charts can become difficult to work with and can even exceed screen resolution. We propose a new visualization approach that affords more scalability and then demonstrate it on systems running with up to 16,384 processes.

In May of 2008, we published online a series of software visualization videos using a method called code_swarm. Shortly thereafter, we made the code open source and its popularity took off. This paper is a study of our code swarm application, comprising its design, results and public response. We share our design methodology, including why we chose the organic information visualization technique, how we designed for both developers and a casual audience, and what lessons we learned from our experiment. We validate the results produced by code_swarm through a qualitative analysis and by gathering online user comments. Furthermore, we successfully released the code as open source, and the software community used it to visualize their own projects and shared their results as well. In the end, we believe code_swarm has positive implications for the future of organic information design and open source information visualization practice.

Though the mediums for visualization are limited, the potential dimensions of a dataset are not. In many areas of scientific study, understanding the correlations between those dimensions and their uncertainties is pivotal to mining useful information from a dataset. Obtaining this insight can necessitate visualizing the many relationships among temporal, spatial, and other dimensionalities of data and its uncertainties. We utilize multiple views for interactive dataset exploration and selection of important features, and we apply those techniques to the unique challenges of cosmological particle datasets. We show how interactivity and incorporation of multiple visualization techniques help overcome the problem of limited visualization dimensions and allow many types of uncertainty to be seen in correlation with other variables.

With the success of open source software projects, such as Apache and Mozilla, comes the opportunity to study the development process. In this paper, we present StarGate: a novel system for visualizing software projects. Whereas previous software project visualizations concentrated mainly on the source code changes, we literally place the developers in the center of our design. Developers are grouped visually into clusters corresponding to the areas of the file repository they work on the most. Connections are drawn between people who communicate via email. The changes to the repository are also displayed. With StarGate, it is easy to look beyond the source code and see trends in developer activity. The system can be used by anyone interested in the project, but it especially benefits project managers, project novices and software engineering researchers.

Abstract graphs or networks are a commonly recurring data type in many fields. In order to visualize such graphs effectively, the graph must be laid out on the screen coherently. Many algorithms exist to do this, but many of these algorithms tend to be very slow when the input graph is large. This paper presents a new approach to the large graph layout problem, which quickly generates an effective layout. This new method proceeds by generating a clustering hierarchy for the graph, applying a treemap to this hierarchy, and finally placing the graph vertices in their associated regions in the treemap. It is ideal for interactive systems where operations such as semantic zooming are to be performed, since most of the work is done in the initial hierarchy calculation, and it takes very little work to recalculate the layout. This method is also valuable in that the resulting layout can be used as the input to an iterative algorithm (e.g., a force directed method), which greatly reduces the number of iterations required to converge to a near optimal layout.

Animation is frequently utilized to visually depict change in time- varying data sets. For this task, it is a natural fit. Yet explicit animation is rarely employed for static data. We discuss the use of animation to overcome three common limitations faced by information visualization applications in the context of small-display devices: constraints on the output display, limited interaction possibilities, and high data density. We provide concrete examples of applying animation to combat such limitations for four common visualization types: geospatial data, treemaps, parallel coordinate displays, and large graphs. Unlike previous work which examines animation for maintaining user orientation during view changes or for displaying data variables, we discuss animation's utility for multiplexing available screen space. In the context of constrained displays, we demonstrate its ability to effectively gain screen resolution, to quickly uncover trends, to help find unexpected data patterns, and to reduce visual clutter.

The widespread use of mobile devices brings opportunities to capture large-scale, continuous information about human behavior. Mobile data has tremendous value, leading to business opportunities, market strategies, security concerns, etc. Visual analytics systems that support interactive exploration and discovery are needed to extracting insight from the data. However, visual analysis of complex social-spatial-temporal mobile data presents several challenges. We have created MobiVis, a visual analytics tool, which incorporates the idea of presenting social and spatial information in one heterogeneous network. The system supports temporal and semantic filtering through an interactive time chart and ontology graph, respectively, such that data subsets of interest can be isolated for close-up investigation. "Behavior rings," a compact radial representation of individual and group behaviors, is introduced to allow easy comparison of behavior patterns. We demonstrate the capability of MobiVis with the results obtained from analyzing the MIT Reality Mining dataset.

The ever-increasing amounts of simulation data produced by scientists demand high-end parallel visualization capability. However, image compositing, which requires interprocessor communication, is often the bottleneck stage for parallel rendering of large volume data sets. Existing image compositing solutions either incur a large number of messages exchanged among processors (such as the direct send method), or limit the number of processors that can be effectively utilized (such as the binary swap method). We introduce a new image compositing algorithm, called 2-3 swap, which combines the flexibility of the direct send method and the optimality of the binary swap method. The 2-3 swap algorithm allows an arbitrary number of processors to be used for compositing, and fully utilizes all participating processors throughout the course of the compositing. We experiment with this image compositing solution on a supercomputer with thousands of processors, and demonstrate its great flexibility as well as scalability.

The visualization of complex 3D images remains a challenge, a fact that is magnified by the difficulty to classify or segment volume data. In this paper, we introduce size-based transfer functions, which map the local scale of features to color and opacity. Features in a data set with similar or identical scalar values can be classified based on their relative size. We achieve this with the use of scale fields, which are 3D fields that represent the relative size of the local feature at each voxel. We present a mechanism for obtaining these scale fields at interactive rates, through a continuous scale-space analysis and a set of detection filters. Through a number of examples, we show that size-based transfer functions can improve classification and enhance volume rendering techniques, such as maximum intensity projection. The ability to classify objects based on local size at interactive rates proves to be a powerful method for complex data exploration.

Network data frequently arises in a wide variety of fields, and node-link diagrams are a very natural and intuitive representation of such data. In order for a node-link diagram to be effective, the nodes must be arranged well on the screen. While many graph layout algorithms exist for this purpose, they often have limitations such as high computational complexity or node colocation. This paper proposes a new approach to graph layout through the use of space filling curves which is very fast and guarantees that there will be no nodes that are colocated. The resulting layout is also aesthetic and satisfies several criteria for graph layout effectiveness.

Quality assessment plays a crucial role in data analysis. In this paper, we present a reduced-reference approach to volume data quality assessment. Our algorithm extracts important statistical information from the original data in the wavelet domain. Using the extracted information as feature and predefined distance functions, we are able to identify and quantify the quality loss in the reduced or distorted version of data, eliminating the need to access the original data. Our feature representation is naturally organized in the form of multiple scales, which facilitates quality evaluation of data with different resolutions. The feature can be effectively compressed in size. We have experimented with our algorithm on scientific and medical data sets of various sizes and characteristics. Our results show that the size of the feature does not increase in proportion to the size of original data. This ensures the scalability of our algorithm and makes it very applicable for quality assessment of large-scale data sets. Additionally, the feature could be used to repair the reduced or distorted data for quality improvement. Finally, our approach can be treated as a new way to evaluate the uncertainty introduced by different versions of data.

The ability to identify and present the most essential aspects of time-varying data is critically important in many areas of science and engineering. This paper introduces an importance-driven approach to time-varying volume data visualization for enhancing that ability. By conducting a block-wise analysis of the data in the joint feature-temporal space, we derive an importance curve for each data block based on the formulation of conditional entropy from information theory. Each curve characterizes the local temporal behavior of the respective block, and clustering the importance curves of all the volume blocks effectively classifies the underlying data. Based on different temporal trends exhibited by importance curves and their clustering results, we suggest several interesting and effective visualization techniques to reveal the important aspects of time-varying data.

We present the design of a scalable parallel pathline construction method for visualizing large time-varying 3D vector fields. A 4D (i.e., time and the 3D spatial domain) representation of the vector field is introduced to make a timeaccurate depiction of the flow field. This representation also allows us to obtain pathlines through streamline tracing in the 4D space. Furthermore, a hierarchical representation of the 4D vector field, constructed by clustering the 4D field, makes possible interactive visualization of the flow field at different levels of abstraction. Based on this hierarchical representation, a data partitioning scheme is designed to achieve high parallel efficiency. We demonstrate the performance of parallel pathline visualization using data sets obtained from terascale flow simulations. This new capability will enable scientists to study their time-varying vector fields at the resolution and interactivity previously unavailable to them. 1.

The dataset generated by a large-scale numerical simulation may include thousands of timesteps and hundreds of variables describing different aspects of the modeled physical phenomena. In order to analyze and understand such data, scientists need the capability to explore simultaneously in the temporal, spatial, and variable domains of the data. Such capability, however, is not generally provided by conventional visualization tools. This paper presents a new visualization interface addressing this problem. The interface consists of three components which abstracts the complexity of exploring in temporal, variable, and spatial domain, respectively. The first component displays time histograms of the data, helps the user identify timesteps of interest, and also helps specify time-varying features. The second component displays correlations between variables in parallel coordinates and enables the user to verify those correlations and possibly identity unanticipated ones. The third component allows the user to more closely explore and validate the data in spatial domain while rendering multiple variables into a single visualization in a user controllable fashion. Each of these three components is not only an interface but is also the visualization itself, thus enabling efficient screen-space usage. The three components are tightly linked to facilitate tri-space data exploration, which offers scientists new power to study their time-varying, multivariate volume data.

For displaying a dense graph, an adjacency matrix is superior than a node-link diagram because it is more compact and free of visual clutter. A node-link diagram, however, is far better for the task of path finding because a path can be easily traced by following the corresponding links, provided that the links are not heavily crossed or tangled.We augment adjacency matrices with path visualization and associated interaction techniques to facilitate path finding. Our design is visually pleasing, and also effectively displays multiple paths based on the design commonly found in metro maps. We illustrate and assess the key aspects of our design with the results obtained from two case studies and an informal user study.

Abstract Paper‐cut is one of the most characteristic Chinese folk arts, often used during festivals and celebrations. Chinese artists have used paper‐cut to make animations. Typical paper‐cut artwork is made with 2D illustrations on paper, and making many frames necessary for an entire animation can be tedious and expensive. We present a system that allows a designer to directly annotate a 3D model with paper‐cut patterns, while still allowing for adding an artistic touch to the design. We have designed special motifs coupled with templates, resulting in a parameterized set of decorative paper‐cut patterns which give the artist flexible control and allow editing of the pattern size, orientation, and shape. The artist chooses a pattern and places the pattern on the model in the desired position. The system determines actual surface coverage and trims the object's surface geometry to simulate the cut‐out effect we observe in traditional 2D paper‐cut art. We demonstrate that our system allows faster and easier addition of paper‐cut decoration to 3D models compared to general purpose modeling tools, such as Maya. Animations made with the 3D paper‐cut models escape from the limitations of traditional 2D paper‐cut animation on the movement in perspective, furthermore, our system allows for easy pattern animations on 3D models, which is very powerful but hard to do with traditional paper‐cut animation. Copyright © 2007 John Wiley & Sons, Ltd.

Hardware-accelerated volume rendering using the GPU is now the standard approach for real-time volume rendering, although limited graphics memory can present a problem when rendering large volume data sets. Volumetric compression in which the decompression is coupled to rendering has been shown to be an effective solution to this problem; however, most existing techniques were developed in the context of software volume rendering, and all but the simplest approaches are prohibitive in a real-time hardware-accelerated volume rendering context. In this paper we present a novel block-based transform coding scheme designed specifically with real-time volume rendering in mind, such that the decompression is fast without sacrificing compression quality. This is made possible by consolidating the inverse transform with dequantization in such a way as to allow most of the reprojection to be precomputed. Furthermore, we take advantage of the freedom afforded by off-line compression in order to optimize the encoding as much as possible while hiding this complexity from the decoder. In this context we develop a new block classification scheme which allows us to preserve perceptually important features in the compression. The result of this work is an asymmetric transform coding scheme that allows very large volumes to be compressed and then decompressed in real-time while rendering on the GPU.

Visualization exploration is the process of extracting insight from data via interaction with visual depictions of that data. Visualization exploration is more than presentation; the interaction with both the data and its depiction is as important as the data and depiction itself. Significant visualization research has focused on the generation of visualizations (the depiction); less effort has focused on the exploratory aspects of visualization (the process). However, without formal models of the process, visualization exploration sessions cannot be fully utilized to assist users and system designers. Toward this end, we introduce the P-Set Model of Visualization Exploration for describing this process and a framework to encapsulate, share, and analyze visual explorations. In addition, systems utilizing the model and framework are more efficient as redundant exploration is avoided. Several examples drawn from visualization applications demonstrate these benefits. Taken together, the model and framework provide an effective means to exploit the information within the visual exploration process.

The output from the massively parallel scientific simulations is so voluminous and complex that advanced visualization technologies are necessary to interpret the calculated results. Even though visualization technology has progressed significantly in recent years, we are barely capable of visualizing and analyzing terascale data to its full extent, and petascale datasets are on the horizon. This workshop aims at addressing this pressing issue by fostering communication between visualization researchers and practitioners. The workshop attendees will be introduced to the latest and greatest research innovations in large data visualization and also help direct further research direction through an open discussion session.

We have developed a novel analytic capability for scientists and engineers to obtain insight from ongoing large-scale parallel unstructured mesh simulations running on thousands of processors. The breakthrough is made possible by a new approach that visualizes partial differential equation (PDE) solution data simultaneously while a parallel PDE solver executes. The solution field is pipelined directly to volume rendering, which is computed in parallel using the same processors that solve the PDE equations. Because our approach avoids the bottlenecks associated with transferring and storing large volumes of output data, it offers a promising approach to overcoming the challenges of visualization of petascale simulations. The submitted video demonstrates real-time on-the-fly monitoring, interpreting, and steering from a remote laptop computer of a 1024-processor simulation of the 1994 Northridge earthquake in Southern California.

Parallel supercomputing has traditionally focused on the inner kernel of scientific simulations: the solver. The front and back ends of the simulation pipeline - problem description and interpretation of the output - have taken a back seat to the solver when it comes to attention paid to scalability and performance, and are often relegated to offline, sequential computation. As the largest simulations move beyond the realm of the terascale and into the petascale, this decomposition in tasks and platforms becomes increasingly untenable. We propose an end-to-end approach in which all simulation components - meshing, partitioning, solver, and visualization - are tightly coupled and execute in parallel with shared data structures and no intermediate I/O. We present our implementation of this new approach in the context of octree-based finite element simulation of earthquake ground motion. Performance evaluation on up to 2048 processors demonstrates the ability of the end-to-end approach to overcome the scalability bottlenecks of the traditional approach

An important challenge in the application of direct volume rendering to time-varying data is the specification of transfer functions for all time steps. Very little research has been devoted to this problem, however. To address this issue we propose an approach which allows simultaneous classification of the entire time series. We explore options for transfer function specification that are based, either directly or indirectly, on the time histogram. Furthermore, we consider how to effectively provide feedback for interactive classification by exploring options for simultaneous rendering of the time series, again based on the time histogram. Finally, we apply this approach to several large time-varying data sets where we show that the important features at all times are captured with about the same effort it takes to classify one time step using conventional classification.

This paper demonstrates the prevalence of a shared characteristic between visualizations and images of nature. We have analyzed visualization competitions and user studies of visualizations and found that the more preferred, better performing visualizations exhibit more natural characteristics. Due to our brain being wired to perceive natural images [SO01], testing a visualization for properties similar to those of natural images can help show how well our brain is capable of absorbing the data. In turn, a metric that finds a visualization’s similarity to a natural image may help determine the effectiveness of that visualization. We have found that the results of comparing the sizes and distribution of the objects in a visualization with those of natural standards strongly correlate to one’s preference of that visualization.

User studies, evaluations, and comparisons of tree visualization systems have so far focused on questions that can readily be answered by simple, automated queries without needing visualization. Studies are lacking on the actual use of tree visualization in discovering intrinsic, hidden, non-trivial and potentially valuable knowledge. We have thus formulated a set of tree exploration tasks not previously considered and have performed user studies and analysis to determine how visualization helps users to perform these tasks. In our study, we evaluated three systems: RINGS (a node-link representation), Treemap (a containment representation), and Windows Explorer. Our findings suggest a few ways that tree visualization helps users to perceive different aspects of hierarchical structured information. We then explain how these visual representations are able to trigger human perception to make these discoveries.

Social network analysis is an active area of study beyond sociology. It uncovers the invisible relationships between actors in a network and provides understanding of social processes and behaviors. It has become an important technique in a variety of application areas such as the Web, organizational studies, and homeland security. This paper presents a visual analytics tool, OntoVis, for understanding large, heterogeneous social networks, in which nodes and links could represent different concepts and relations, respectively. These concepts and relations are related through an ontology (also known as a schema). OntoVis is named such because it uses information in the ontology associated with a social network to semantically prune a large, heterogeneous network. In addition to semantic abstraction, OntoVis also allows users to do structural abstraction and importance filtering to make large networks manageable and to facilitate analytic reasoning. All these unique capabilities of OntoVis are illustrated with several case studies.

Terascale simulations produce data that is vast in spatial, temporal, and variable domains, creating a formidable challenge for subsequent analysis. Feature extraction as a data reduction method offers a viable solution to this large data problem. This paper presents a new approach to the problem of extracting and visualizing 4D features within large volume data. Conventional methods requires either an analytical description of the feature of interest or tedious manual intervention throughout the feature extraction and tracking process. We show that it is possible for a visualization system to "learn" to extract and track features in complex 4D flow field according to their "visual" properties, location, shape, and size. The basic approach is to employ machine learning in the process of visualization. Such an intelligent system approach is powerful because it allows us to extract and track an feature of interest in a high-dimensional space without explicitly specifying the relations between those dimensions, resulting in a greatly simplified and intuitive visualization interface.

Rendering directly from packed or compressed volume data formats using graphics hardware has advantages in terms of memory consumption and bandwidth, but results in lower-quality images due to the prohibitive cost of performing interpolation and gradient-based shading on the reconstructed data. The problem with the existing method lies in its close coupling of decompression and interpolation. We demonstrate that deferred filtering overcomes this problem by using a two-pass decompression and rendering strategy. With this method interpolation and gradient calculations are very efficient, allowing high quality rendering directly from packed or compressed volume data. We evaluate the cost of creating interpolated, gradient-shaded renderings using traditional on-the-fly decompression and deferred filtering, and show that deferred filtering can provide up to twenty times speed-up for high quality rendering.

Artificial neural networks are computer software or hardware models inspired by the structure and behavior of neurons in the human nervous system. As a powerful learning tool, increasingly neural networks have been adopted by many large-scale information processing applications but there is no a set of well defined criteria for choosing a neural network. The user mostly treats a neural network as a black box and cannot explain how learning from input data was done nor how performance can be consistently ensured. We have experimented with several information visualization designs aiming to open the black box to possibly uncover underlying dependencies between the input data and the output data of a neural network. In this paper, we present our designs and show that the visualizations not only help us design more efficient neural networks, but also assist us in the process of using neural networks for problem solving such as performing a classification task.

Many methods have been developed for monitoring network traffic, both using visualization and statistics. Most of these methods focus on the detection of suspicious or malicious activities. But what they often fail to do refine and exercise measures that contribute to the characterization of such activities and their sources, once they are detected. In particular, many tools exist that detect network scans or visualize them at a high level, but not very many tools exist that are capable of categorizing and analyzing network scans. This paper presents a means of facilitating the process of characterization by using visualization and statistics techniques to analyze the patterns found in the timing of network scans through a method of continuous improvement in measures that serve to separate the components of interest in the characterization so the user can control separately for the effects of attack tool employed, performance characteristics of the attack platform, and the effects of network routing in the arrival patterns of hostile probes. The end result is a system that allows large numbers of network scans to be rapidly compared and subsequently identified.

THIS issue contains extended versions of three of the bestrated application papers taken from the IEEE Visualization 2004 Conference. Application papers present the contribution of visualization techniques toward the understanding of application-specific data. In particular, application papers must explain the effectiveness of the visualization methods for the particular application domain. For the IEEE Visualization 2004 Conference, 24 out of 71 submitted applications papers were chosen for inclusion in the conference program. As paper cochairs of the IEEE Visualization applications track, we started from the opinion of the reviewers. From the 24 accepted papers, we have chosen three very high quality submissions and asked the authors to provide extended and revised versions of their original work for this issue. Each of the extended application papers was thoroughly reviewed by several experts and underwent revisions before it was recommended for acceptance. We are grateful to the reviewers for their detailed and timely reviews and to David Ebert, Editor-in-Chief of TVCG, for his strong support and assistance. The paper “Reconstruction and Visualization of Planetary Nebulae” by Marcus Magnor, Gordon Kindlmann, Charles Hansen, and Neb Duric exploits strong symmetry characteristics of planetary nebulae to recover spatial structures from 2D images. With GPU-based volume rendering and nonlinear optimization, the nebula’s emission density is recovered. Fascinating 3D visualizations of planetary nebulae are created, which further educational purposes and can support astrophysicists in better understanding the formation process of these phenomena. The paper “Advanced Virtual Endoscopic Pituitary Surgery” by Andre Neubauer, Stefan Wolfsberger, MarieTherese Forster, Lukas Mroz, Rainer Wegenkittl, and Katja Buhler investigates a minimally invasive endoscopic procedure for tumor removal. STEPS, a virtual endoscopy system, is presented to train surgeons on the transsphenoidal approach and assist experienced surgeons in planning a real endoscopic intervention. Interactive visualization is achieved through first-hit ray casting. The application provides various navigation and perception aids in the simulation of the surgical procedure. The paper “Visualization of Geologic Stress Perturbations Using Mohr Diagrams” by Patricia Crossno, David H. Rogers, Rebecca M. Brannon, David Coblentz, and Joanne T. Fredrich presents an inspection tool for finite element analyses of 3D real-valued second-order tensors. The Mohr diagram is a paper and pencil method from the material mechanics community. The authors originate the adaptation of this diagram type to visualize perturbed in-situ stress fields of geologic features. The Mohr diagram is used as an interactive glyph for probing as well as filtering through brushing and linking of principal stresses. Scientific visualization research is largely application driven. Many techniques are strongly motivated by the application scenario for which they were developed. Visualization research cannot be a goal by itself. It just acts as an interface tool to the domain expert to facilitate his understanding of his data and problems. Applications are therefore crucial in providing fruitful incitation for the further research of the visualization experts. The papers presented here are prime examples of how visualization helps to understand intricate application phenomena. We hope that the readers will find the papers inspiring and stimulating samples of work from our fascinating field.

In volume data visualization, the classification step is used to determine voxel visibility and is usually carried out through the interactive editing of a transfer function that defines a mapping between voxel value and color/opacity. This approach is limited by the difficulties in working effectively in the transfer function space beyond two dimensions. We present a new approach to the volume classification problem which couples machine learning and a painting metaphor to allow more sophisticated classification in an intuitive manner. The user works in the volume data space by directly painting on sample slices of the volume and the painted voxels are used in an iterative training process. The trained system can then classify the entire volume. Both classification and rendering can be hardware accelerated, providing immediate visual feedback as painting progresses. Such an intelligent system approach enables the user to perform classification in a much higher dimensional space without explicitly specifying the mapping for every dimension used. Furthermore, the trained system for one data set may be reused to classify other data sets with similar characteristics.

This paper presents a parallel visualization pipeline implemented at the Pittsburgh Supercomputing Center (PSC) for studying the largest earthquake simulation ever performed. The simulation employs 100 million hexahedral cells to model 3D seismic wave propagation of the 1994 Northridge earthquake. The time-varying dataset produced by the simulation requires terabytes of storage space. Our solution for visualizing such terascale simulations is based on a parallel adaptive rendering algorithm coupled with a new parallel I/O strategy which effectively reduces interframe delay by dedicating some processors to I/O and preprocessing tasks. In addition, a 2D vector field visualization method and a 3D enhancement technique are incorporated into the parallel visualization framework to help scientists better understand the wave propagation both on and under the ground surface. Our test results on the HP/Compaq AlphaServer operated at the PSC show that we can completely remove the I/O bottlenecks commonly present in time-varying data visualization. The high-performance visualization solution we provide to the scientists allows them to explore their data in the temporal, spatial, and variable domains at high resolution. The new high-resolution explorability, likely not available to most computational science groups, will help lead to many new insights.

The continuing advancement of plasma science is central to realizing fusion as an inexpensive and safe energy source. Gryokinetic simulations of plasmas are fundamental to the understanding of turbulent transport in fusion plasma. This work discusses the visualization challenges presented by gyrokinetic simulations using magnetic field line following coordinates, and presents an effective solution exploiting programmable graphics hardware to enable interactive volume visualization of 3D plasma flow on a toroidal coordinate system. The new visualization capability can help scientists better understand three-dimensional structures of the modeled phenomena. Both the limitations and future promise of the hardware-accelerated approach are also discussed.

An important task in volume rendering is the visualization of boundaries between materials. This is typically accomplished using transfer functions that increase opacity based on a voxel's value and gradient. Lighting also plays a crucial role in illustrating surfaces. In this paper we present a multi-dimensional transfer function method for enhancing surfaces, not through the variation of opacity, but through the modification of surface shading. The technique uses a lighting transfer function that takes into account the distribution of values along a material boundary and features a novel interface for visualizing and specifying these transfer functions. With our method, the user is given a means of visualizing boundaries without modifying opacity, allowing opacity to be used for illustrating the thickness of homogeneous materials through the absorption of light.

Many large scale physics-based simulations which take place on PC clusters or supercomputers produce huge amounts of data including vector fields. While these vector data such as electromagnetic fields, fluid flow fields, or particle paths can be represented by lines, the sheer number of the lines overwhelms the memory and computation capability of a high-end PC used for visualization. Further, very dense or intertwined lines, rendered with traditional visualization techniques, can produce unintelligible results with unclear depth relationships between the lines and no sense of global structure. Our approach is to apply a lighting model to the lines and sample them into an anisotropic voxel representation based on spherical harmonics as a preprocessing step. Then we evaluate and render these voxels for a given view using traditional volume rendering. For extremely large line based datasets, conversion to anisotropic voxels reduces the overall storage and rendering for O(n) lines to O(1) with a large constant that is still small enough to allow meaningful visualization of the entire dataset at nearly interactive rates on a single commodity PC.

Most visualizations of security-related network data require large amounts of finely detailed, high-dimensional data. However, in some cases, the data available can only be coarsely detailed because of security concerns or other limitations. How can interesting security events still be discovered in data that lacks important details, such as IP addresses, network security alarms, and labels? In this paper, we discuss a system we have designed that takes very coarsely detailed data--basic, summarized information of the activity on each TCP port during each given hour--and uses visualization to help uncover interesting security events.

The security of Internet routing is a major concern because attacks and errors can result in data packets not reaching their intended destination and/or falling into the wrong hands. A key step in improving routing security is to analyze and understand it. In the past, we and other researchers have presented various visual-based, statistical-based, and signature-based methods of analyzing Internet routing data.

Decision trees are commonly used for classification. We propose to use decision trees not just for classification but also for the wider purpose of knowledge discovery, because visualizing the decision tree can reveal much valuable information in the data. We introduce PaintingClass, a system for interactive construction, visualization and exploration of decision trees. PaintingClass provides an intuitive layout and convenient navigation of the decision tree. PaintingClass also provides the user the means to interactively construct the decision tree. Each node in the decision tree is displayed as a visual projection of the data. Through actual examples and comparison with other classification methods, we show that the user can effectively use PaintingClass to construct a decision tree and explore the decision tree to gain additional knowledge.

Interactive data visualization is inherently an iterative trial-and-error process searching for an ideal set of parameters for classifying and rendering features of interest in the data. This paper presents 3-d region growing based techniques that can assist the users to locate and define features of interest in volume data more quickly and more accurately. One technique employs partial region growing to generate a 2-d transfer function that effectively reveals the full features of interest. The other technique uses the result of full region growing to systematically construct a boundary surface for the extracted features. The resulting polygonal representation of the boundary surface can facilitate comparison, measurement, and simulation. A visual assessment method is suggested by using the extracted volume and surface information. These techniques either shorten or completely eliminate the typical trial-and-error step in the process of interactive data exploration.

This paper presents a parallel adaptive rendering algorithm and its performance for visualizing time-varying unstructured volume data generated from large-scale earthquake simulations. The objective is to visualize 3D seismic wave propagation generated from a 0.5 Hz simulation of the Northridge earthquake, which is the highest resolution volume visualization of an earthquake simulation performed to date. This scalable high-fidelity visualization solution we provide to the scientists allows them to explore in the temporal, spatial, and visualization domain of their data at high resolution. This new high resolution explorability, likely not presently available to most computational science groups, will help lead to many new insights. The performance study we have conducted on a massively parallel computer operated at the Pittsburgh Supercomputing Center helps direct our design of a simulation-time visualization strategy for the higher-resolution, 1Hz and 2 Hz, simulations.

Classification operations in a data-mining task are often performed using decision trees. The visual-based approach to decision tree construction has gained increasing popularity. We present StarClass, a new interactive visual classification method. This method maps multi-dimensional data to the visual display using star coordinates, allowing the user to interact with the display to create a decision tree. Preliminary evaluation indicates that this new technique is as effective as state-of-the-art algorithmic classification methods, and more effective than the previous visual-based methods. StarClass also offers additional advantages such as improving the user's understanding of the data.

No abstract available.

This paper describes a set of techniques developed for the visualization of high-resolution volume data generated from industrial computed tomography for nondestructive testing (NDT) applications. Because the data are typically noisy and contain fine features, direct volume rendering methods do not always give us satisfactory results. We have coupled region growing techniques and a 2D histogram interface to facilitate volumetric feature extraction. The new interface allows the user to conveniently identify, separate or composite, and compare features in the data. To lower the cost of segmentation, we show how partial region growing results can suggest a reasonably good classification function for the rendering of the whole volume. The NDT applications that we work on demand visualization tasks including not only feature extraction and visual inspection, but also modeling and measurement of concealed structures in volumetric objects. An efficient filtering and modeling process for generating surface representation of extracted features is also introduced. Four CT data sets for preliminary NDT are used to demonstrate the effectiveness of the new visualization strategy that we have developed.

The Internet pervades many aspects of our lives and is becoming indispensable to critical functions in areas such as commerce, government, production and general information dissemination. To maintain the stability and efficiency of the Internet, every effort must be made to protect it against various forms of attacks, malicious users, and errors. A key component in the Internet security effort is the routine examination of Internet routing data, which unfortunately can be too large and complicated to browse directly. We have developed an interactive visualization process which proves to be very effective for the analysis of Internet routing data. In this application paper, we show how each step in the visualization process helps direct the analysis and glean insights from the data. These insights include the discovery of patterns, detection of faults and abnormal events, understanding of event correlations, formation of causation hypotheses, and classification of anomalies. We also discuss lessons learned in our visual analysis study.

In the traditional volume visualization paradigm, the user specifies a transfer function that assigns each scalar value to a color and opacity by defining an opacity and a color map function. The transfer function has two limitations. First, the user must define curves based on histogram and value rather than seeing and working with the volume itself. Second, the transfer function is inflexible in classifying regions of interest, where values at a voxel such as intensity and gradient are used to differentiate material, not talking into account additional properties such as texture and position. We describe an intuitive user interface for specifying the classification functions that consists of the users painting directly on sample slices of the volume. These painted regions are used to automatically define high-dimensional classification functions that can be implemented in hardware for interactive rendering. The classification of the volume is iteratively improved as the user paints samples, allowing intuitive and efficient viewing of materials of interest.

In this paper, we present a novel visualization technique-kinetic visualization-that uses motion along a surface to aid in the perception of 3D shape and structure of static objects. The method uses particle systems, with rules such that particles flow over the surface of an object to not only bring out, but also attract attention to information on a shape that might not be readily visible with a conventional rendering method which uses lighting and view changes. Replacing still images with animations in this fashion, we demonstrate with both surface and volumetric models in the accompanying videos that, in many cases, the resulting visualizations effectively enhance the perception of three-dimensional shape and structure. We also describe how, for both types of data, a texture-based representation of this motion can be used for interactive visualization using PC graphics hardware. Finally, the results of a user study that we have conducted are presented, which show evidence that the supplemental motion cues can be helpful.

This paper presents a study of field line visualization techniques. To address both the computational and perceptual issues in visualizing large scale, complex, dense field line data commonly found in many scientific applications, a new texture-based field line representation which we call self-orienting surfaces is introduced This scalable representation facilitates hardware-accelerated rendering and incorporation of various perceptually-effective techniques, resulting in intuitive visualization and interpretation of the data under study. An electromagnetic data set obtained from accelerator modeling and a fluid flow data set from aerodynamics modeling are used for evaluation and demonstration of the techniques.

This paper presents a set of feature enhancement techniques. coupled with hardware-accelerated non-photorealistic rendering for generating more perceptually effective visualizations of multidimensional, multivariate volume data, such as those obtained from typical computational fluid dynamics simulations. For time-invariant data, one or more variables are used to either highlight important features in another variable, or add contextural information to the visualization. For time-varying data, rendering of each time step also takes into account the values at neighboring time steps to reinforce the perception of the changing features in the data over time. With hardware-accelerated rendering, interactive visualization becomes possible leading to increased explorability and comprehension of the data.

This paper introduces an interactive classification technique for volume data, called ISpace, which uses Independent Component Analysis (ICA) and a multidimensional histogram of the volume data in a transformed space. Essentially, classification in the volume domain becomes equivalent to interactive clipping in the ICA space, which as demonstrated using several examples is more intuitive and direct for the user to classify data. The result is an opacity transfer function defined for rendering multivariate scalar volume data.

This paper presents two new hardware-assisted rendering techniques developed for interactive visualization of the terascale data generated from numerical modeling of next-generation accelerator designs. The first technique, based on a hybrid rendering approach, makes possible interactive exploration of large-scale particle data from particle beam dynamics modeling. The second technique, based on a compact texture-enhanced representation, exploits the advanced features of commodity graphics cards to achieve perceptually effective visualization of the very dense and complex electromagnetic fields produced from the modeling of reflection and transmission properties of open structures in an accelerator design. Because of the collaborative nature of the overall accelerator modeling project, the visualization technology developed is for both desktop and remote visualization settings. We have tested the techniques using both time-varying particle data sets containing up to one billion particles per time step and electromagnetic field data sets with millions of mesh elements.

The current state of the art in visualization research places strong emphasis on different techniques to derive insight from disparate types of data. However, little work has investigated the visualization process itself. The information content of the visualization process - the results, history, and relationships between those results - is addressed by this work. A characterization of the visualization process is discussed, leading to a general model of the visualization exploration process. The model, based upon a new parameter derivation calculus, can be used for automated reporting, analysis, or visualized directly. An XML-based language for expressing visualization sessions using the model is also described. These sessions can then be shared and reused by collaborators. The model, along with the XML representation, provides an effective means to utilize information within the visualization process to further data exploration.

Motion provides strong visual cues for the perception of shape and depth, as demonstrated by cognitive scientists and visual artists. This paper presents a novel visualization technique-kinetic visualization -that uses particle systems to add supplemental motion cues which can aid in the perception of shape and spatial relationships of static objects. Based on a set of rules following perceptual and physical principles, particles flowing over the surface of an object not only bring out, but also attract attention to, essential information on the shape of the object that might not be readily visible with conventional rendering that uses lighting and view changes. Replacing still images with animations in this fashion, we demonstrate with both surface and volumetric models in the accompanying videos that in many cases the resulting visualizations effectively enhance the perception of three-dimensional shape and structure. The results of a preliminary user study that we have conducted also show evidence that the supplemental motion cues help.

Internet connectivity is defined by a set of routing protocols which let the routers that comprise the Internet backbone choose the best route for a packet to reach its destination. One way to improve the security and performance of Internet is to routinely examine the routing data. In this case study, we show how interactive visualization of Border Gateway Protocol (BGP) data helps characterize routing behavior, identify weaknesses in connectivity which could potentially cripple the Internet, as well as detect and explain actual anomalous events.

We present a scalable volume rendering technique that exploits lossy compression and low-cost commodity hardware to permit highly interactive exploration of time-varying scalar volume data. A palette-based decoding technique and an adaptive bit allocation scheme are developed to fully utilize the texturing capability of a commodity 3D graphics card. Using a single PC equipped with a modest amount of memory, a texture-capable graphics card and an inexpensive disk array, we are able to render hundreds of time steps of regularly gridded volume data (up to 42 million voxels each time step) at interactive rates. By clustering multiple PCs together, we demonstrate the data-size scalability of our method. The frame rates achieved make possible the interactive exploration of data in the temporal, spatial and transfer function domains. A comprehensive evaluation of our method based on experimental studies using data sets (up to 134 million voxels per time step) from turbulence flow simulations is also presented.

The authors present a watercolor inspired method for the rendering of surfaces. Our approach mimics the watercolor process by building up an illuminated scene through the compositing of several layers of semitransparent paint. The key steps consist of creating textures for each layer using LIC (Line Integral Convolution) of Perlin Noise (K. Perlin, 1985), and then calculating the layer thickness distribution using an inverted subtractive lighting model. The resulting watercolor-style images have color coherence that results from the mixing of a limited palette of paints. The new lighting model helps to better convey large shape changes, while texture orientations give hints of less dominant features. The rendered images therefore possess perceptual clues to more effectively communicate shape and texture information.

To seek a low-cost, extensible solution for the large-scale data visualization problem, a visual computing system is designed as a result of a collaboration between industry and government research laboratories in Japan, also with participation by researchers in U.S. This scalable system is a commodity PC cluster equipped with the VolumePro 500 volume graphics cards and a specially designed image compositing hardware. Our performance study shows such a system is capable of interactive rendering 5123 and 10243 volume data and highly scalable. In particular, with such a system, simulation and visualization can be performed concurrently which allows scientists to monitor and tune their simulations on the fly. In this paper, both the system and hardware designs are presented.

In this paper we present a hardware-assisted rendering technique coupled with a compression scheme for the interactive visual exploration of time-varying scalar volume data. A palette-based decoding technique and an adaptive bit allocation scheme are developed to fully utilize the texturing capability of a commodity 3-D graphics card. Using a single PC equipped with a modest amount of memory, a texture capable graphics card, and an inexpensive disk array, we are able to render hundreds of time steps of regularly gridded volume data (up to 45 millions voxels each time step) at interactive rates, permitting the visual exploration of large scientific data sets in both the temporal and spatial domain.

Exploring complex, very large data sets requires interfaces to present and navigate through the visualization of the data. Two types of audience benefit from such coherent organization and representation: first, the user of the visualization system can examine and evaluate their data more efficiently; second, collaborators or reviewers can quickly understand and extend the visualization. The needs of these two groups are addressed by the spreadsheet-like interface described in this paper. The interface represents a 2D window in a multidimensional visualization parameter space. Data is explored by navigating this space via the interface. The visualization space is presented to the user in a manner that clearly identifies which parameters correspond to which visualized result. Operations defined on this space can be applied which generate new parameters or results. Combined with a general-purpose interpreter, these functions can be utilized to quickly extract desired results. Finally, by encapsulating the visualization process, redundant exploration is eliminated and collaboration is facilitated. The efficacy of this novel interface is demonstrated through examples using a variety of data sets in different domains.

This paper presents an end-to-end, low-cost solution for visualizing time-varying volume data rendered on a parallel computer located at a remote site. Pipelining and careful grouping of processors are used to hide I/O time and to maximize processors utilization. Compression is used to significantly cut down the cost of transferring output images from the parallel computer to a display device through a widearea network. This complete rendering pipeline makes possible highly efficient rendering and remote viewing of high resolution time-varying data sets in the absence of high-speed network and parallel I/O support. To study the performance of this rendering pipeline and to demonstrate high-performance remote visualization, tests were conducted on a PC cluster in Japan as well as an SGI Origin 2000 operated at the NASA Ames Research Center with the display located at UC Davis.

As the size and complexity of data sets continues to increase, the development of user interfaces and interaction techniques that expedite the process of exploring that data must receive new attention. Regardless of the speed of rendering, it is important to coherently organize the visual process of exploration: this information both grants insights about the data to a user and can be used by collaborators to understand the results. To fulfil these needs, we present a spreadsheet-like interface to data exploration. The interface displays a 2-dimensional window into visualization parameter space which users manipulate as they search for desired results. Through tabular organization and a clear correspondence between parameters and results, the interface eases the discovery, comparison and analysis of the underlying data. Users can utilize operators and the integrated interpreter to further explore and automate the visualization process; using a method introduced in this paper, these operations can be applied to cells in different stacks of the interface. Via illustrations using a variety of data sets, we demonstrate the efficacy of this novel interface.

We demonstrate the use of a combination of perceptually effective techniques for visualizing magnetic field data from the DIII-D Tokamak. These techniques can be implemented to run very efficiently on machines with hardware support for OpenGL. Interactive speeds facilitate clear communication of magnetic field structure, enhancing fusion scientists' understanding of their data, and thereby accelerating their research.

For types of data visualization where the cost of producing images is high, and the relationship between the rendering parameters and the image produced is less than obvious, a visual representation of the exploration process can make the process more efficient and effective. Image graphs represent not only the results but also the process of data visualization. Each node in an image graph consists of an image and the corresponding visualization parameters used to produce it. Each edge in a graph shows the change in rendering parameters between the two nodes it connects. Image graphs are not just static representations; users can interact with a graph to review a previous visualization session or to perform new rendering. Operations which cause changes in rendering parameters can propagate through the graph. The user can take advantage of the information in image graphs to understand how certain parameter changes affect visualization results. Users can also share image graphs to streamline the process of collaborative visualization. We have implemented a volume visualization system using the image graph interface, and the examples presented come from this application.

We present a fast volume rendering algorithm for time-varying fields. We propose a new data structure, called time-space partitioning (TSP) tree, that can effectively capture both the spatial and the temporal coherence from a time-varying field. Using the proposed data structure, the rendering speed is substantially improved. In addition, our data structure helps to maintain the memory access locality and to provide the sparse data traversal so that our algorithm becomes suitable for large-scale out-of-core applications. Finally, our algorithm allows flexible error control for both the temporal and the spatial coherence so that a trade-off between image quality and rendering speed is possible. We demonstrate the utility and speed of our algorithm with data from several time-varying CFD simulations. Our rendering algorithm can achieve substantial speedup while the storage space overhead for the TSP tree is kept at a minimum.

The paper discusses techniques for extracting feature lines from three-dimensional unstructured grids. The twin objectives are to facilitate the interactive manipulation of these typically very large and dense meshes, and to clarify the visualization of the solution data that accompanies them. The authors describe the perceptual importance of specific viewpoint-dependent and view-independent features, discuss the relative advantages and disadvantages of several alternative algorithms for identifying these features (taking into consideration both local and global criteria), and demonstrate the results of these methods on a variety of different data sets.

This paper presents an out-of-core approach for interactive streamline construction on large unstructured tetrahedral meshes containing millions of elements. The out-of-core algorithm uses an octree to partition and restructure the raw data into subsets stored into disk files for fast data retrieval. A memory management policy tailored to the streamline calculations is used such that, during the streamline construction, only a very small amount of data are brought into the main memory on demand. By carefully scheduling computation and data fetching, the overhead of reading data from the disk is significantly reduced and good memory performance results. This out-of-core algorithm makes possible interactive streamline visualization of large unstructured-grid data sets on a single mid-range workstation with relatively low main-memory capacity: 5-15 megabytes. We also demonstrate that this approach is much more efficient than relying on virtual memory and operating system's paging algorithms.

Streamline construction is one of the most fundamental techniques for visualizing steady flow fields. Streamribbons and streamtubes are extensions for visualizing the rotation and the expansion of the flow. The paper presents efficient algorithms for constructing streamlines, streamribbons, and streamtubes on unstructured grids. A specialized Runge-Kutta method is developed to speed up the tracing of streamlines. Explicit solutions are derived for calculating the angular rotation rates of streamribbons and the radii of streamtubes. In order to simplify mathematical formulations and reduce computational costs, all calculations are carried out in the canonical coordinate system instead of the physical coordinate system. The resulting speed up in overall performance helps explore large flow fields.

The plotting of streamlines is an effective way of visualizing fluid motion in steady flows. Additional information about the flowfield, such as local rotation and expansion, can be shown by drawing in the form of a ribbon or tube. In this paper, we present efficient algorithms for the construction of streamlines, streamribbons and streamtubes on unstructured grids. A specialized version of the Runge-Kutta method has been developed to speed up the integration of particle paths. We have also derived closed-form solutions for calculating angular rotation rate and radius to construct streamribbons and streamtubes, respectively. According to our analysis and test results, these formulations are two to four times better in performance than previous numerical methods. As a large number of traces are calculated, the improved performance could be significant.

Air flowing around the wing tips of an airplane forms horizontal tornado-like vortices that can be dangerous to following aircraft. The dynamics of such vortices, including ground and atmospheric effects, can be predicted by numerical simulation, allowing the safety and capacity of airports to be improved. We introduce three-dimensional techniques for visualizing time-dependent, two-dimensional wake vortex computations, and the hazard strength of such vortices near the ground. We describe a vortex core tracing algorithm and a local tiling method to visualize the vortex evolution. The tiling method converts time-dependent, two-dimensional vortex cores into three-dimensional vortex tubes. Finally, a novel approach is used to calculate the induced rolling moment on the following airplane at each grid point within a region near the vortex tubes and thus allows three-dimensional visualization of the hazard strength of the vortices.>

The paper describes a highly interactive method for computer visualization of simultaneous three-dimensional vector and scalar flow fields in convection-diffusion systems. This method allows a computational fluid dynamics user to visualize the basic physical process of dispersion and mixing rather than just the vector and scalar values computed by the simulation. It is based on transforming the vector field from a traditionally Eulerian reference frame into a Lagrangian reference frame. Fluid elements are traced through the vector field for the mean path as well as the statistical dispersion of the fluid elements about the mean position by using added scalar information about the root mean square value of the vector field and its Lagrangian time scale. In this way, clouds of fluid elements are traced not just mean paths. We have used this method to visualize the simulation of an industrial incinerator to help identify mechanisms for poor mixing.>

Recent advances in software and hardware technology have made direct ray-traced volume rendering of 3-d scalar data a feasible and effective method for imaging of the data's contents.The time costs of these rendering techniques still do not permit full interaction with the data, and all of the parameters effecting the resulting images.This paper presents a set of real-time interaction techniques which have been developed to permit exploration of a volume data set.Within the limitation of a static viewpoint, the user is able to interactively alter the position and shape of an area of interest, and modify local viewing parameters.A run length encoded cache of volume rendering samples provides the means to rerender the volume at interactive rates.The user locates and plants "seeds" in areas of interest through the use of data slicing and isosurface techniques.Image processing techniques applied to volumes ( i.e. volume processing), can then automatically form regions of interest which in turn modify the rendering parameters.This "region growing" of "seedlings" incrementally alters the image in real-time providing further visual cues concerning the contents of the data.These tools allow interactive exploration of internal structures in the data which may be obscured by other imaging algorithms.Magnetic Resonance Angiography (MRA) provides a driving application for this technology.Results from preliminary studies of MRA data are included.1

A technique is given for computer visualization of simultaneous three-dimensional vector and scalar fields such as velocity and temperature in reacting fluid flow fields. The technique, which is called Virtual Smoke, simulates the use of colored smoke for experimental gaseous fluid flow visualization. However, it is noninvasive and can animate, in particular, the dynamic behaviors of steady-state or instantaneous flow fields obtained from numerical simulations. Virtual Smoke is based on volume seeds and volume seedlings, which are direct volume visualization methods previously developed for highly interactive scalar volume data exploration. Data from combustion simulations are used to demonstrate the effectiveness of Virtual Smoke.>

Abstract Ray‐traced volume rendering has been shown to be an effective method for visualizing 3D scalar data. However, with currently available workstation technology, interactive volume exploration using conventional volume rendering is still too slow to be attractive. This paper describes an enhanced volume rendering method which allows interactive changes of rendering parameters such as colour and opacity maps. An innovative technique is provided which allows the user to plant a ‘seed’ in the volume to rapidly modify local shading parameters. For a fixed viewing position, the user can interactively explore specific regions of interest. Furthermore, a virtual cutting technique with the exploratory seed allows the user to remove surfaces and see the internal structure of the volume. Examples demonstrate these techniques as an attractive option in many applications.

Abstract Most current Common Lisp compilers generate more efficient code when supplied with data type information. However, in keeping with standard Lisp programming style, most programmers are reluctant to provide type information; they simply allow the run‐time type system to manage the data types accordingly. To fill this gap, we have designed and implemented a type inference system for Common Lisp (TICL). TICL takes a Lisp program that has been annotated with a few type declarations, adds as many declarations as possible, and produces a type declared program. The compiler can then use this information to generate more efficient code. Measurements indicate that a 20 per cent speed improvement can generally be achieved.