Research in Explainable AI (XAI) has shown that explanations can improve users’ understanding of AI models, improve user performance and potentially reduce overreliance on AI predictions. However, this is mostly evaluated by static rather than dynamic measures, and the role of XAI on learning over trials is rarely studied. In this study, we use a context-free sequence prediction task, in which 458 participants predict the next symbol in a fixed sequence (with some noise) over 80 trials. We compare performance with AI and XAI advice against no AI support, and subsequently we test for learning by taking away the AI support after 40 trials (i.e., a reversal study design). Our results show that users learn faster with XAI than with AI without explanations or no AI and are better able to recover in performance from the removal of AI. However, the benefits of XAI on learning are much smaller for more difficult tasks. This work demonstrates the benefits of repeated measures user studies and multilevel modeling to better understand learning processes in XAI. It also shows the potential of AI explanations to help users to learn and poses XAI design suggestions to support learning in human-AI collaboration.

Abstract Local surrogate learning is a popular and successful method for machine learning explanation. It uses synthetic transfer data to approximate a complex reference model. The sampling technique used for this transfer data has a significant impact on the provided explanation, but remains relatively unexplored in literature. In this work, we explore alternative sampling techniques in pursuit of more faithful and robust explanations, and present LEMON: a sampling technique that samples directly from the desired distribution instead of reweighting samples as done in other explanation techniques (e.g., LIME). Next, we evaluate our technique in a synthetic and UCI dataset-based experiment, and show that our sampling technique yields more faithful explanations compared to current state-of-the-art explainers.

Businesses in high-risk environments have been reluctant to adopt modern machine learning approaches due to their complex and uninterpretable nature. Most current solutions provide local, instance-level explanations, but this is insufficient for understanding the model as a whole. In this work, we show that strategy clusters (i.e., groups of data instances that are treated distinctly by the model) can be used to understand the global behavior of a complex ML model. To support effective exploration and understanding of these clusters, we introduce StrategyAtlas, a system designed to analyze and explain model strategies. Furthermore, it supports multiple ways to utilize these strategies for simplifying and improving the reference model. In collaboration with a large insurance company, we present a use case in automatic insurance acceptance, and show how professional data scientists were enabled to understand a complex model and improve the production model based on these insights.

Recent advances in computational and algorithmic power are evolving the field of medical imaging rapidly. In cancer research, many new directions are sought to characterize patients with additional imaging features derived from radiology and pathology images. The emerging field of Computational Pathology targets the high-throughput extraction and analysis of the spatial distribution of cells from digital histopathology images. The associated morphological and architectural features allow researchers to quantify and characterize new imaging biomarkers for cancer diagnosis, prognosis, and treatment decisions. However, while the image feature space grows, exploration and analysis become more difficult and ineffective. There is a need for dedicated interfaces for interactive data manipulation and visual analysis of computational pathology and clinical data. For this purpose, we present IIComPath, a visual analytics approach that enables clinical researchers to formulate hypotheses and create computational pathology pipelines involving cohort construction, spatial analysis of image-derived features, and cohort analysis. We demonstrate our approach through use cases that investigate the prognostic value of current diagnostic features and new computational pathology biomarkers.

In this paper, we introduce 11-20 (Image Insight 2020), a multimedia analytics approach for analytic categorization of image collections. Advanced visualizations for image collections exist, but they need tight integration with a machine model to support the task of analytic categorization. Directly employing computer vision and interactive learning techniques gravitates towards search. Analytic categorization, however, is not machine classification (the difference between the two is called the pragmatic gap): a human adds/redefines/deletes categories of relevance on the fly to build insight, whereas the machine classifier is rigid and non-adaptive. Analytic categorization that truly brings the user to insight requires a flexible machine model that allows dynamic sliding on the exploration-search axis, as well as semantic interactions: a human thinks about image data mostly in semantic terms. 11-20 brings three major contributions to multimedia analytics on image collections and towards closing the pragmatic gap. Firstly, a new machine model that closely follows the user's interactions and dynamically models her categories of relevance. II-20's machine model, in addition to matching and exceeding the state of the art's ability to produce relevant suggestions, allows the user to dynamically slide on the exploration-search axis without any additional input from her side. Secondly, the dynamic, 1-image-at-a-time Tetris metaphor that synergizes with the model. It allows a well-trained model to analyze the collection by itself with minimal interaction from the user and complements the classic grid metaphor. Thirdly, the fast-forward interaction, allowing the user to harness the model to quickly expand ("fast-forward") the categories of relevance, expands the multimedia analytics semantic interaction dictionary. Automated experiments show that II-20's machine model outperforms the existing state of the art and also demonstrate the Tetris metaphor's analytic quality. User studies further confirm that II-20 is an intuitive, efficient, and effective multimedia analytics tool.

Machine learning models often exhibit complex behavior that is difficult to understand. Recent research in explainable AI has produced promising techniques to explain the inner workings of such models using feature contribution vectors. These vectors are helpful in a wide variety of applications. However, there are many parameters involved in this process and determining which settings are best is difficult due to the subjective nature of evaluating interpretability. To this end, we introduce EXPLAINEXPLORE: an interactive explanation system to explore explanations that fit the subjective preference of data scientists. We leverage the domain knowledge of the data scientist to find optimal parameter settings and instance perturbations, and enable the discussion of the model and its explanation with domain experts. We present a use case on a real-world dataset to demonstrate the effectiveness of our approach for the exploration and tuning of machine learning explanations.

The field of explainable artificial intelligence aims to help experts understand complex machine learning models. One key approach is to show the impact of a feature on the model prediction. This helps experts to verify and validate the predictions the model provides. However, many challenges remain open. For example, due to the subjective nature of interpretability, a strict definition of concepts such as the contribution of a feature remains elusive. Different techniques have varying underlying assumptions, which can cause inconsistent and conflicting views. In this work, we introduce Local and Global Contribution-Value plots as a novel approach to visualize feature impact on predictions and the relationship with feature value. We discuss design decisions, and show an exemplary visual analytics implementation that provides new insights into the model.

We propose a new visualization of family relations, called family metro map, which is inspired by the famous London Underground Map. A family is represented by a metro line, where the parents are the end nodes and the children the intermediate nodes. We introduce family trees as rooted binary trees and consider the problem of automatically drawing a family metro map for a given family tree without crossing line segments, where the successive nodes of a metro line are placed at equal distances. The inter-node distance may be chosen to depend on the generation of the family, such that metro lines of earlier generations are shown more prominently than metro lines of later generations.

While analyzing multiple data sequences, the following questions typically arise: how does a single sequence change over time, how do multiple sequences compare within a period, and how does such comparison change over time. This paper presents a visual technique named STBins to answer these questions. STBins is designed for visual tracking of individual data sequences and also for comparison of sequences. The latter is done by showing the similarity of sequences within temporal windows. A perception study is conducted to examine the readability of alternative visual designs based on sequence tracking and comparison tasks. Also, two case studies based on real-world datasets are presented in detail to demonstrate usage of our technique.

Abstract The usage of deep learning models for tagging input data has increased over the past years because of their accuracy and high‐performance. A successful application is to score sleep stages. In this scenario, models are trained to predict the sleep stages of individuals. Although their predictive accuracy is high, there are still mis classifications that prevent doctors from properly diagnosing sleep‐related disorders. This paper presents a system that allows users to explore the output of deep learning models in a real‐life scenario to spot and analyze faulty predictions. These can be corrected by users to generate a sequence of sleep stages to be examined by doctors. Our approach addresses a real‐life scenario with absence of ground truth. It differs from others in that our goal is not to improve the model itself, but to correct the predictions it provides. We demonstrate that our approach is effective in identifying faulty predictions and helping users to fix them in the proposed use case.

Abstract Event sequences and time series are widely recorded in many application domains; examples are stock market prices, electronic health records, server operation and performance logs. Common goals for recording are monitoring, root cause analysis and predictive analytics. Current analysis methods generally focus on the exploration of either event sequences or time series. However, deeper insights are gained by combining both. We present a visual analytics approach where users can explore both time series and event data simultaneously, combining visualization, automated methods and human interaction. We enable users to iteratively refine the visualization. Correlations between event sequences and time series can be found by means of an interactive algorithm, which also computes the presence of monotonic effects. We illustrate the effectiveness of our method by applying it to real world and synthetic data sets.

We present RegressionExplorer, a Visual Analytics tool for the interactive exploration of logistic regression models. Our application domain is Clinical Biostatistics, where models are derived from patient data with the aim to obtain clinically meaningful insights and consequences. Development and interpretation of a proper model requires domain expertise and insight into model characteristics. Because of time constraints, often a limited number of candidate models is evaluated. RegressionExplorer enables experts to quickly generate, evaluate, and compare many different models, taking the workflow for model development as starting point. Global patterns in parameter values of candidate models can be explored effectively. In addition, experts are enabled to compare candidate models across multiple subpopulations. The insights obtained can be used to formulate new hypotheses or to steer model development. The effectiveness of the tool is demonstrated for two uses cases: prediction of a cardiac conduction disorder in patients after receiving a heart valve implant and prediction of hypernatremia in critically ill patients.

Forensic analysis of malware activity in network environments is a necessary yet very costly and time consuming part of incident response. Vast amounts of data need to be screened, in a very labor-intensive process, looking for signs indicating how the malware at hand behaves inside e.g., a corporate network. We believe that data reduction and visualization techniques can assist security analysts in studying behavioral patterns in network traffic samples (e.g., PCAP). We argue that the discovery of patterns in this traffic can help us to quickly understand how intrusive behavior such as malware activity unfolds and distinguishes itself from the rest of the traffic.In this paper we present a case study of the visual analytics tool EventPad and illustrate how it is used to gain quick insights in the analysis of PCAP traffic using rules, aggregations, and selections. We show the effectiveness of the tool on real-world data sets involving office traffic and ransomware activity.

Multivariate event sequences are ubiquitous: travel history, telecommunication conversations, and server logs are some examples. Besides standard properties such as type and timestamp, events often have other associated multivariate data. Current exploration and analysis methods either focus on the temporal analysis of a single attribute or the structural analysis of the multivariate data only. We present an approach where users can explore event sequences at multivariate and sequential level simultaneously by interactively defining a set of rewrite rules using multivariate regular expressions. Users can store resulting patterns as new types of events or attributes to interactively enrich or simplify event sequences for further investigation. In Eventpad we provide a bottom-up glyph-oriented approach for multivariate event sequence analysis by searching, clustering, and aligning them according to newly defined domain specific properties. We illustrate the effectiveness of our approach with real-world data sets including telecommunication traffic and hospital treatments.

Abstract Digital image collections contain a wealth of information, which for instance can be used to trace illegal activities and investigate criminal networks. We present a method that enables analysts to reveal relations among people, based on the patterns in their collections. Similar temporal and spatial patterns can be found using a parameterized algorithm, visualization is used to choose the right parameters and to inspect the patterns found. The visualization shows relations between image properties: the person it belongs to, the concepts in the image, its time stamp and location. We demonstrate the method with image collections of 10, 000 people containing 460, 000 images in total.

We present a new approach for the analysis of large image collections. We argue that categorization plays an important role in this process, not only to label images as end result, but also during exploration. Furthermore, to increase the effectiveness and efficiency of the categorization process we enable the use of all available metadata, treated as multivariate data. We identified images, attributes, and categories as important aspects and integrated these in a system called ICLIC. The system consists of four views that are connected by the selection of images, which is a central action in the approach. By using a minimalist interface with only standard metaphors, users are enabled to use the system in short time. The system enables complex queries in a natural way, and it can deal with collections containing more than 100,000 images and more than 1,000 metadata attributes. This was confirmed by two evaluation cycles with domain experts.

Understanding huge multimedia collections is a huge challenge. Given a set of hundreds of thousands or millions of images, how to to understand its contents and how to find the images that are relevant for the task at hand? Using a combination of automated methods, visualization, and interaction, known as visual analytics, is probably the only way to go, combining the strengths of man and machine. An overview is given of trends in data visualization and visual analytics is given, and examples of recent work in multimedia analytics are presented. Exploiting meta-data, using interaction with relatively simple visual representations, and alignment with the work flow of users are promising routes, but scalability and evaluation are still challenging.

For the protection of critical infrastructures against complex virus attacks, automated network traffic analysis and deep packet inspection are unavoidable. However, even with the use of network intrusion detection systems, the number of alerts is still too large to analyze manually. In addition, the discovery of domain-specific multi stage viruses (e.g., Advanced Persistent Threats) are typically not captured by a single alert. The result is that security experts are overloaded with low-level technical alerts where they must look for the presence of an APT. In this paper we propose an alert-oriented visual analytics approach for the exploration of network traffic content in multiple contexts. In our approach CoNTA (Contextual analysis of Network Traffic Alerts), experts are supported to discover threats in large alert collections through interactive exploration using selections and attributes of interest. Tight integration between machine learning and visualization enables experts to quickly drill down into the alert collection and report false alerts back to the intrusion detection system. Finally, we show the effectiveness of the approach by applying it on real world and artificial data sets.

Abstract The exploration of high‐dimensional data is challenging because humans have difficulty to understand more than three dimensions. We present a new visualization concept that enables users to explore such data and, specifically, to learn about important items and features that are unknown or overlooked, based on the items and features that are already known. The visualization consists of two juxtaposed tables: an IF‐Table, showing all items with a selection of features; and an FI‐Table, showing all features with a selection of items. This enables the user to limit the number of visible items and features to those needed for the exploration. The interaction is kept simple: each selection of items and features results in a complete overview of similar and relevant items and features.

We propose a visual analytics approach for the exploration and analysis of dynamic networks. We consider snapshots of the network as points in high-dimensional space and project these to two dimensions for visualization and interaction using two juxtaposed views: one for showing a snapshot and one for showing the evolution of the network. With this approach users are enabled to detect stable states, recurring states, outlier topologies, and gain knowledge about the transitions between states and the network evolution in general. The components of our approach are discretization, vectorization and normalization, dimensionality reduction, and visualization and interaction, which are discussed in detail. The effectiveness of the approach is shown by applying it to artificial and real-world dynamic networks.

Visualization of the trajectories of moving objects leads to dense and cluttered images, which hinders exploration and understanding. It also hinders adding additional visual information, such as direction, and makes it difficult to interactively extract traffic flows, i.e., subsets of trajectories. In this paper we present our approach to visualize traffic flows and provide interaction tools to support their exploration. We show an overview of the traffic using a density map. The directions of traffic flows are visualized using a particle system on top of the density map. The user can extract traffic flows using a novel selection widget that allows for the intuitive selection of an area, and filtering on a range of directions and any additional attributes. Using simple, visual set expressions, the user can construct more complicated selections. The dynamic behaviors of selected flows may then be shown in annotation windows in which they can be interactively explored and compared. We validate our approach through use cases where we explore and analyze the temporal behavior of aircraft and vessel trajectories, e.g., landing and takeoff sequences, or the evolution of flight route density. The aircraft use cases have been developed and validated in collaboration with domain experts.

Most network traffic analysis applications are designed to discover malicious activity by only relying on high-level flow-based message properties. However, to detect security breaches that are specifically designed to target one network (e.g., Advanced Persistent Threats), deep packet inspection and anomaly detection are indispensible. In this paper, we focus on how we can support experts in discovering whether anomalies at message level imply a security risk at network level. In SNAPS (Semantic Network traffic Analysis through Projection and Selection), we provide a bottom-up pixel-oriented approach for network traffic analysis where the expert starts with low-level anomalies and iteratively gains insight in higher level events through the creation of multiple selections of interest in parallel. The tight integration between visualization and machine learning enables the expert to iteratively refine anomaly scores, making the approach suitable for both post-traffic analysis and online monitoring tasks. To illustrate the effectiveness of this approach, we present example explorations on two real-world data sets for the detection and understanding of potential Advanced Persistent Threats in progress.

Abstract In safety and security domains where objects of interest (OOI), such as people, vessels, or transactions, are continuously monitored, automated reasoning is required due to their sheer number and volume of information. We present a method to visually explain the rationale of a reasoning engine that raises an alarm if a certain situation is reached. Based both on evidence from heterogeneous and possibly unreliable sources, and on a domain specific reasoning structure, this engine concludes with a certain probability that, e.g., the OOI is suspected of smuggling. To support decision making, we visualize the rationale, an abstraction of the complicated reasoning structure. The evidence is displayed in a color‐coded matrix that easily reveals if and where observations contradict. In it, domain and operational experts can quickly understand and find complicated patterns and relate them to real‐world situations. Also, two groups of these experts evaluate our system through maritime use cases based on real data.

In moving object visualization, objects and their attributes are commonly represented by glyphs on a geographic map. In areas on the map densely populated by these objects, visual clutter and occlusion of glyphs occur. We propose a method to solve this problem by partitioning the set of all objects into subsets that are each visualized using an aggregated multivariate glyph that shows the distribution of several attributes of its objects, such as heading, type and velocity. We choose the combination of subsets and glyph design such that the glyphs do not overlap and the number of subsets is approximately maximal. The partition is maintained and updated while the objects move. We use examples from the maritime domain, but our method is applicable to a wider range of dynamic data. Through a user study we find that, for a set of representative tasks, our method does not perform significantly worse than competitive visualizations with respect to correctness. Furthermore, it performs significantly better for density comparison tasks in high density data sets. We also find that the participants of the user study have a preference for our method.

Abstract Medicine prescriptions play an important role in medical treatments. More insight in medicine prescription behavior can lead to more efficient and effective treatments, as well as reflection on prescription behavior for specific physicians, types of medicines, or classes of patients. Most current medical visualization systems show health data only from the perspective of patients, whereas to understand prescription behavior multiple perspectives are relevant. We present a new approach to visualize prescription data from four different perspectives: physician, patient, medicine, and prescription. Information about physicians, patients, and medicines is shown in three tables; relations between selected items in these tables are shown using custom glyphs and histograms. These tables can also be used to define selections of prescriptions which can be compared to each other by showing a variety of metrics. This enables physicians and possibly other stakeholders to perform a wide variety of queries and inspections, while the use of familiar metaphors, such as tables and histograms, enables them to use the system in short time. This was confirmed by an evaluation session with six neurologists from an institute of epileptology. Our system is tailored to medicine prescription data, but we argue that the underlying pattern in the data is ubiquitous, and that hence our approach can be useful for many other cases where A provides B to C.

We present a visualization method for the interactive exploration of predicted positions of moving objects, in particular, ocean-faring vessels. Two simple prediction models, one based on similarity to historical trajectories and one on Monte Carlo simulation, are presented. The prediction models generate temporal probability density fields starting from a known situation. We use contours to visualize spatio-temporal zones of these density fields. Predictions are split into a configurable number of segments for which we render one or more contours. Users, investigating and exploring the possible development of a situation, can see where a vessel will be in the near future according to a given prediction model. Through a number of real-world use cases and a discussion with users, we show our methods can be used in monitoring traffic for collision avoidance, and detecting illegal activities, like piracy or smuggling. By applying our methods to pedestrian movements, we show that our methods can also be applied to a different domain.

A common task in visualization is to quickly find interesting items in large sets. When appropriate metadata is missing, automatic queries are impossible and users have to inspect all elements visually. We compared two fundamentally different, but obvious display modes for this task and investigated the difference with respect to effectiveness, efficiency, and satisfaction. The static mode is based on the page metaphor and presents successive pages with a static grid of items. The moving mode is based on the conveyor belt metaphor and lets a grid of items slide though the screen in a continuous flow. In our evaluation, we applied both modes to the common task of browsing images. We performed two experiments where 18 participants had to search for certain target images in a large image collection. The number of shown images per second (pace) was predefined in the first experiment, and under user control in the second one. We conclude that at a fixed pace, the mode has no significant impact on the recall. The perceived pace is generally slower for moving mode, which causes users to systematically choose for a faster real pace than in static mode at the cost of recall, keeping the average number of target images found per second equal for both modes.

Networks are present in many fields such as finance, sociology, and transportation. Often these networks are dynamic: they have a structural as well as a temporal aspect. In addition to relations occurring over time, node information is frequently present such as hierarchical structure or time-series data. We present a technique that extends the Massive Sequence View ( msv) for the analysis of temporal and structural aspects of dynamic networks. Using features in the data as well as Gestalt principles in the visualization such as closure, proximity, and similarity, we developed node reordering strategies for the msv to make these features stand out that optionally take the hierarchical node structure into account. This enables users to find temporal properties such as trends, counter trends, periodicity, temporal shifts, and anomalies in the network as well as structural properties such as communities and stars. We introduce the circular msv that further reduces visual clutter. In addition, the (circular) msv is extended to also convey time-series data associated with the nodes. This enables users to analyze complex correlations between edge occurrence and node attribute changes. We show the effectiveness of the reordering methods on both synthetic and a rich real-world dynamic network data set.

Network data is ubiquitous; e-mail traffic between persons, telecommunication, transport and financial networks are some examples. Often these networks are large and multivariate, besides the topological structure of the network, multivariate data on the nodes and links is available. Currently, exploration and analysis methods are focused on a single aspect; the network topology or the multivariate data. In addition, tools and techniques are highly domain specific and require expert knowledge. We focus on the non-expert user and propose a novel solution for multivariate network exploration and analysis that tightly couples structural and multivariate analysis. In short, we go from Detail to Overview via Selections and Aggregations (DOSA): users are enabled to gain insights through the creation of selections of interest (manually or automatically), and producing high-level, infographic-style overviews simultaneously. Finally, we present example explorations on real-world datasets that demonstrate the effectiveness of our method for the exploration and understanding of multivariate networks where presentation of findings comes for free.

A regular map is a symmetric tiling of a closed surface, in the sense that all faces, vertices, and edges are topologically indistinguishable. Platonic solids are prime examples, but also for surfaces with higher genus such regular maps exist. We present a new method to visualize regular maps. Space models are produced by matching regular maps with target shapes in the hyperbolic plane. The approach is an extension of our earlier work. Here a wider variety of target shapes is considered, obtained by duplicating spherical and toroidal regular maps, merging triangles, punching holes, and gluing the edges. The method produces about 45 new examples, including the genus 7 Hurwitz surface.

Networks are present in many fields such as finance, sociology, and transportation. Often these networks are dynamic: they have a structural as well as a temporal aspect. We present a technique that extends the Massive Sequence View (MSV) for the analysis of the temporal and structural aspects of dynamic networks. Using features in the data as well as in the visualization based on the Gestalt principles closure, proximity, and similarity, we developed node reordering strategies for the MSV to make these features stand out. This enables users to find temporal properties such as trends, counter trends, periodicity, temporal shifts, and anomalies in the network as well as structural properties such as communities and stars. We show the effectiveness of the reordering methods on both synthetic and real-world transaction data sets.

The visualization group of TU/e has worked on information visualization since 1998. In this talk I will give an overview of our work on tree, graph, and multivariate visualization, for a variety of applications, including software visualization. Techniques like cushion treemaps, squarified treemaps, hierarchical edge bundles, and flexible linked axes will be illustrated with demos. Furthermore, I will reflect on approaches for the development of new presentations and dealing with evaluation, based on our experience and lessons we learned.

Abstract We present a novel visual exploration method based on small multiples and large singles for effective and efficient data analysis. Users are enabled to explore the state space by offering multiple alternatives from the current state. Users can then select the alternative of choice and continue the analysis. Furthermore, the intermediate steps in the exploration process are preserved and can be revisited and adapted using an intuitive navigation mechanism based on the well‐known undo‐redo stack and filmstrip metaphor. As proof of concept the exploration method is implemented in a prototype. The effectiveness of the exploration method is tested using a formal user study comparing four different interaction methods. By using Small Multiples as data exploration method users need fewer steps in answering questions and also explore a significantly larger part of the state space in the same amount of time, providing them with a broader perspective on the data, hence lowering the chance of missing important features. Also, users prefer visual exploration with small multiples over non‐small multiple variants.

Abstract Julia sets based on quadratic polynomials have a very simple definition, yet a highly intricate shape. Our contribution is to provide a visual explanation for this complexity. To this end we show the construction of Julia sets as a dynamic process, in contrast to showing just a static image of the set itself. Our method is based on the Inverse Iteration Method (IIM). We start with a disk, which is successively distorted. The crucial step is to show an animation of the effect of taking a root of a subset of the complex plane. We present four different approaches for this, using a Riemann surface, a corkscrew, a fan, and disks as metaphors. We packaged our results in an interactive tool with a simple interface, such that everybody can view and inspect these for different Julia sets. The results are useful for teaching complex analysis, promoting mathematics, entertainment, and, above all, as a visual explanation for the complexity of Julia sets.

We present a novel technique-Compressed Adjacency Matrices-for visualizing gene regulatory networks. These directed networks have strong structural characteristics: out-degrees with a scale-free distribution, in-degrees bound by a low maximum, and few and small cycles. Standard visualization techniques, such as node-link diagrams and adjacency matrices, are impeded by these network characteristics. The scale-free distribution of out-degrees causes a high number of intersecting edges in node-link diagrams. Adjacency matrices become space-inefficient due to the low in-degrees and the resulting sparse network. Compressed adjacency matrices, however, exploit these structural characteristics. By cutting open and rearranging an adjacency matrix, we achieve a compact and neatly-arranged visualization. Compressed adjacency matrices allow for easy detection of subnetworks with a specific structure, so-called motifs, which provide important knowledge about gene regulatory networks to domain experts. We summarize motifs commonly referred to in the literature, and relate them to network analysis tasks common to the visualization domain. We show that a user can easily find the important motifs in compressed adjacency matrices, and that this is hard in standard adjacency matrix and node-link diagrams. We also demonstrate that interaction techniques for standard adjacency matrices can be used for our compressed variant. These techniques include rearrangement clustering, highlighting, and filtering.

We present the results of a study comparing five directed-edge representations for use in 2D, screen-based node-link diagrams. The goal of this work is to extend the understanding of tradeoffs and best practices for the representation of edges in directed graphs and to help practitioners choose among different options. Our work applies to graphs in which directed links are depicted using lines connecting the nodes. We tested five different edge representations chosen carefully based on user feedback to thoroughly cover the directed-edge design space. We also investigated how the use of pattern compression affects performance and subjective user preference. The article presents detailed results regarding the significant performance and preference differences between directed-edge representations and provides practical recommendations on their use.

We present a method to interactively explore multiple attributes in trajectory data using density maps, i.e., images that show an aggregate overview of massive amounts of data. So far, density maps have mainly been used to visualize single attributes. Density maps are created in a two-way procedure; first smoothed trajectories are aggregated in a density field, and then the density field is visualized. In our approach, the user can explore attributes along trajectories by calculating a density field for multiple subsets of the data. These density fields are then either combined into a new density field or first visualized and then combined. Using a widget, called a distribution map, the user can interactively define subsets in an effective and intuitive way, and, supported by high-end graphics hardware the user gets fast feedback for these computationally expensive density field calculations. We show the versatility of our method with use cases in the maritime domain: to distinguish between periods in the temporal aggregation, to find anomalously behaving vessels, to solve ambiguities in density maps via drill down in the data, and for risk assessments. Given the generic framework and the lack of domain-specific assumptions, we expect our concept to be applicable for trajectories in other domains as well.

Abstract We propose visual analytics techniques to support concurrent comparison of hundreds of cumulatively weighted instances of a single hierarchy. This includes a node‐link representation of the hierarchy where nodes depict the weights of all instances with high‐density heat maps that are grouped and aligned to ease cross‐referencing. Hierarchy exploration is facilitated by smoothly animated expansion and collapse of its branches. Detailed information about hierarchy structure, weights, and meta‐data is provided by secondary linked visualizations. These techniques have been implemented in a prototype tool, in which the computational analysis concerns have been strictly separated from the visualization concerns. The analysis algorithms are extensible via a script engine. We discuss the effectiveness of our techniques for the visual analytic process of microbe community profiling experts.

Abstract The analysis of large graphs plays a prominent role in various fields of research and is relevant in many important application areas. Effective visual analysis of graphs requires appropriate visual presentations in combination with respective user interaction facilities and algorithmic graph analysis methods. How to design appropriate graph analysis systems depends on many factors, including the type of graph describing the data, the analytical task at hand and the applicability of graph analysis methods. The most recent surveys of graph visualization and navigation techniques cover techniques that had been introduced until 2000 or concentrate only on graph layouts published until 2002. Recently, new techniques have been developed covering a broader range of graph types, such as time‐varying graphs. Also, in accordance with ever growing amounts of graph‐structured data becoming available, the inclusion of algorithmic graph analysis and interaction techniques becomes increasingly important. In this State‐of‐the‐Art Report, we survey available techniques for the visual analysis of large graphs. Our review first considers graph visualization techniques according to the type of graphs supported. The visualization techniques form the basis for the presentation of interaction approaches suitable for visual graph exploration. As an important component of visual graph analysis, we discuss various graph algorithmic aspects useful for the different stages of the visual graph analysis process. We also present main open research challenges in this field.

Abstract There are many visualizations that show the trajectory of a moving object to obtain insights in its behavior. In this user study, we test the performance of three of these visualizations with respect to three movement features that occur in vessel behavior. Our goal is to compare the recently presented vessel density by Willems et al. [ WvdWvW09 ] with well‐known trajectory visualizations such as an animation of moving dots and the space‐time cube. We test these visualizations with common maritime analysis tasks by investigating the ability of users to find stopping objects, fast moving objects, and estimate the busiest routes in vessel trajectories. We test the robustness of the visualizations towards scalability and the influence of complex trajectories using small‐scale synthetic data sets. The performance is measured in terms of correctness and response time. The user test shows that each visualization type excels for correctness for a specific movement feature. Vessel density performs best for finding stopping objects, but does not perform significantly less than the remaining visualizations for the other features. Therefore, vessel density is a nice extension in the toolkit for analyzing trajectories of moving objects, in particular for vessel movements, since stops can be visualized better, and the performance for comparing lanes and finding fast movers is at a similar level as established trajectory visualizations.

Multivariate data visualization is a classic topic, for which many solutions have been proposed, each with its own strengths and weaknesses. In standard solutions the structure of the visualization is fixed, we explore how to give the user more freedom to define visualizations. Our new approach is based on the usage of Flexible Linked Axes: The user is enabled to define a visualization by drawing and linking axes on a canvas. Each axis has an associated attribute and range, which can be adapted. Links between pairs of axes are used to show data in either scatter plot- or Parallel Coordinates Plot-style. Flexible Linked Axes enable users to define a wide variety of different visualizations. These include standard methods, such as scatter plot matrices, radar charts, and PCPs [11]; less well known approaches, such as Hyperboxes [1], TimeWheels [17], and many-to-many relational parallel coordinate displays [14]; and also custom visualizations, consisting of combinations of scatter plots and PCPs. Furthermore, our method allows users to define composite visualizations that automatically support brushing and linking. We have discussed our approach with ten prospective users, who found the concept easy to understand and highly promising.

We consider moving objects as multivariate time-series. By visually analyzing the attributes, patterns may appear that explain why certain movements have occurred. Density maps as proposed by Scheepens et al. [25] are a way to reveal these patterns by means of aggregations of filtered subsets of trajectories. Since filtering is often not sufficient for analysts to express their domain knowledge, we propose to use expressions instead. We present a flexible architecture for density maps to enable custom, versatile exploration using multiple density fields. The flexibility comes from a script, depicted in this paper as a block diagram, which defines an advanced computation of a density field. We define six different types of blocks to create, compose, and enhance trajectories or density fields. Blocks are customized by means of expressions that allow the analyst to model domain knowledge. The versatility of our architecture is demonstrated with several maritime use cases developed with domain experts. Our approach is expected to be useful for the analysis of objects in other domains.

No abstract.

Symbols are used in scatterplots to encode data in a way that is appropriate for perception through human visual channels. Color is believed to be the most dominant channel with lightness regarded as the most important one of three color dimensions. We study lightness perception in scatterplots in the context of analytic tasks requiring symbol discrimination. More specifically, we performed an experiment to measure human performance in three visual analytic tasks. Outlined circles and unframed spots, equally sized, with a uniform luminance that was varied at ten or eleven equispaced levels between black and white were used as symbols and displayed on a uniform white background. Sixteen subjects divided in two groups, participated in the experiment and their task performance times were recorded. We propose a model to describe the process. The perception of lightness is assumed to be an early step in the complex cognitive process to mediate discrimination, and psychophysical laws are used to describe this perceptual mapping. Different mapping schemes are compared by regression on the experimental data. The results show that approximate homogeneity of lightness perception exists in our complex tasks and can be closely described by a blended combination of two opposite power functions assuming either the light end or the dark end of the lightness scale as the starting point. The model further yields discriminability scales of lightness for sparse scatterplots with a white background.

Greetings from the program committee of Pacific Visualization 2010. This was the third year of the conference. The quality of this year's program conveys exciting work that is being done in many areas of visualization. In the few years since the inception of the meeting, there has been a steady increase in the number of submissions and their overall technical strength. This reflects the maturity of the field, and perhaps increasing support for a spring meeting to complement VisWeek held in the fall.

Symbols are used in scatterplots to encode data in a way that is appropriate for perception through human visual channels. Symbol size is believed to be the second dominant channel after color. We study symbol size perception in scatterplots in the context of analytic tasks requiring size discrimination. More specifically, we performed an experiment to measure human performance in three visual analytic tasks. Circles are used as the representative symbol, with eight, linearly varying radii; 24 persons, divided across three groups, participated; and both objective and subjective measures were obtained. We propose a model to describe the results. The perception of size is assumed to be an early step in the complex cognitive process to mediate discrimination, and psychophysical laws are used to describe this perceptual mapping. Different mapping schemes are compared by regression on the experimental data. The results show that approximate homogeneity of size perception exists in our complex tasks and can be closely described by a power law transformation with an exponent of 0.4. This yields an optimal scale for symbol size discrimination.

Abstract Parallel coordinate plots (PCPs) are a well‐known visualization technique for viewing multivariate data. In the past, various visual modifications to PCPs have been proposed to facilitate tasks such as correlation and cluster identification, to reduce visual clutter, and to increase their information throughput. Most modifications pertain to the use of color and opacity, smooth curves, or the use of animation. Although many of these seem valid improvements, only few user studies have been performed to investigate this, especially with respect to cluster identification. We performed a user study to evaluate cluster identification performance – with respect to response time and correctness – of nine PCP variations, including standard PCPs. To generate the variations, we focused on covering existing techniques as well as possible while keeping testing feasible. This was done by adapting and merging techniques, which led to the following novel variations. The first is an effective way of embedding scatter plots into PCPs. The second is a technique for highlighting fuzzy clusters based on neighborhood density. The third is a spline‐based drawing technique to reduce ambiguity. The last is a pair of animation schemes for PCP rotation. We present an overview of the tested PCP variations and the results of our study. The most important result is that a fair number of the seemingly valid improvements, with the exception of scatter plots embedded into PCPs, do not result in significant performance gains.

Abstract In interactive visualization, selection techniques such as dynamic queries and brushing are used to specify and extract items of interest. In other words, users define areas of interest in data space that often have a clear semantic meaning. We call such areas Semantic Zones, and argue that support for their manipulation and reasoning with them is highly useful during exploratory analysis. An important use case is the use of these zones across different subsets of the data, for instance to study the population of semantic zones over time. To support this, we present the Select & Slice Table. Semantic zones are arranged along one axis of the table, and data subsets are arranged along the other axis of the table. Each cell contains a set of items of interest from a data subset that matches the selection specifications of a zone. Items in cells can be visualized in various ways, as a count, as an aggregation of a measure, or as a separate visualization, such that the table gives an overview of the relationship between zones and data subsets. Furthermore, users can reuse zones, combine zones, and compare and trace items of interest across different semantic zones and data subsets. We present two case studies to illustrate the support offered by the Select & Slice table during exploratory analysis of multivariate data.

Symbols are frequently used to represent data objects in visualization. An appropriate contrast between symbols is a precondition that determines the efficiency of a visual analysis process. We study the contrast between different types of symbols in the context of scatterplots, based on user testing and a quantitative model for symbol contrast. In total, 32 different symbols were generated by using four sizes, two classes (polygon-and asterisk shaped), and four categories of rotational symmetry; and used three different tasks. From the user test results an internal separation space is established for the symbol types under study. In this space, every symbol is represented by a point, and the visual contrasts defined by task performance between the symbols are represented by the distances between the points. The positions of the points in the space, obtained by multidimensional scaling (MDS), reveal the effects of different visual feature scales. Also, larger distances imply better symbol separation for visual tasks, and therefore indicate appropriate choices for symbols. The resulting configurations are discussed, and a number of patterns in the relation between properties of the symbols and the resulting contrast are identified. In short we found that the size effect in the space is not linear and more dominant than shape effect.

Visualization aims at providing insight to its users. Now and then I am a user myself, and use visualization trying to solve a puzzle and to satisfy my curiosity. Simple questions turn out to be challenging problems, leading to a personal quest for their solution and resulting in intriguing images and animations. In my presentation I will present three such puzzles, all in the area of mathematical visualization.

Graphs are often visualized using node-link representations: vertices are depicted as dots, edges are depicted as (poly)lines connecting two vertices. A directed edge running from vertex A to B is generally visualized using an arrow representation: a (poly)line with a triangular arrowhead at vertex B. Although this representation is intuitive, it is not guaranteed that a user is able to determine edge direction as quickly and unambiguously as possible; alternative representations that exhibit less occlusion and visual clutter might be better suited. To investigate this, we developed five additional directed-edge representations using combinations of shape and color. We performed a user study in which subjects performed different tasks on a collection of graphs using these representations and combinations thereof to investigate which representation is best in terms of speed and accuracy. We present our initial hypotheses, the outcome of the user studies, and recommendations regarding directed-edge visualization.

Abstract Graphs depicted as node‐link diagrams are widely used to show relationships between entities. However, node‐link diagrams comprised of a large number of nodes and edges often suffer from visual clutter. The use of edge bundling remedies this and reveals high‐level edge patterns. Previous methods require the graph to contain a hierarchy for this, or they construct a control mesh to guide the edge bundling process, which often results in bundles that show considerable variation in curvature along the overall bundle direction. We present a new edge bundling method that uses a self‐organizing approach to bundling in which edges are modeled as flexible springs that can attract each other. In contrast to previous methods, no hierarchy is used and no control mesh. The resulting bundled graphs show significant clutter reduction and clearly visible high‐level edge patterns. Curvature variation is furthermore minimized, resulting in smooth bundles that are easy to follow. Finally, we present a rendering technique that can be used to emphasize the bundling.

Abstract We propose a geographical visualization to support operators of coastal surveillance systems and decision making analysts to get insights in vessel movements. For a possibly unknown area, they want to know where significant maritime areas, like highways and anchoring zones, are located. We show these features as an overlay on a map. As source data we use AIS data: Many vessels are currently equipped with advanced GPS devices that frequently sample the state of the vessels and broadcast them. Our visualization is based on density fields that are derived from convolution of the dynamic vessel positions with a kernel. The density fields are shown as illuminated height maps. Combination of two fields, with a large and small kernel provides overview and detail. A large kernel provides an overview of area usage revealing vessel highways. Details of speed variations of individual vessels are shown with a small kernel, highlighting anchoring zones where multiple vessels stop. Besides for maritime applications we expect that this approach is useful for the visualization of moving object data in general.

A regular map is a tiling of a closed surface into faces, bounded by edges that join pairs of vertices, such that these elements exhibit a maximal symmetry. For genus 0 and 1 (spheres and tori) it is well known how to generate and present regular maps, the Platonic solids are a familiar example. We present a method for the generation of space models of regular maps for genus 2 and higher. The method is based on a generalization of the method for tori. Shapes with the proper genus are derived from regular maps by tubification: edges are replaced by tubes. Tessellations are produced using group theory and hyperbolic geometry. The main results are a generic procedure to produce such tilings, and a collection of intriguing shapes and images. Furthermore, we show how to produce shapes of genus 2 and higher with a highly regular structure.

We present a new method for the visual analysis of multivariate system traces. Our method combines three perspectives: (1) a schematic diagram, (2) time series plots and (3) a state transition graph. After we show how these perspectives are related, we discuss their integration into a single solution for the visual analysis of multivariate system traces. The combination of the three perspectives provides the user with a rich analysis interface that enables gaining significant insight into system behavior. We illustrate some of the advantages of our approach by providing a real-world use case.

This paper presents a new information visualization framework that supports the analytical reasoning process. It consists of three views - a data view, a knowledge view and a navigation view. The data view offers interactive information visualization tools. The knowledge view enables the analyst to record analysis artifacts such as findings, hypotheses and so on. The navigation view provides an overview of the exploration process by capturing the visualization states automatically. An analysis artifact recorded in the knowledge view can be linked to a visualization state in the navigation view. The analyst can revisit a visualization state from both the navigation and knowledge views to review the analysis and reuse it to look for alternate views. The whole analysis process can be saved along with the synthesized information. We present a user study and discuss the perceived usefulness of a prototype based on this framework that we have developed.

Abstract We provide a novel visualization method for the comparison of hierarchically organized data. Our technique visualizes a pair of hierarchies that are to be compared and simultaneously depicts how these hierarchies are related by explicitly visualizing the relations between matching subhierarchies. Elements that are unique to each hierarchy are shown, as well as the way in which hierarchy elements are relocated, split or joined. The relations between hierarchy elements are visualized using Hierarchical Edge Bundles (HEBs). HEBs reduce visual clutter, they visually emphasize the aforementioned splits, joins, and relocations of subhierarchies, and they provide an intuitive way in which users can interact with the relations. The focus throughout this paper is on the comparison of different versions of hierarchically organized software systems, but the technique is applicable to other kinds of hierarchical data as well. Various data sets of actual software systems are used to show how our technique can be employed to easily spot splits, joins, and relocations of elements, how sorting both hierarchies with respect to each other facilitates comparison tasks, and how user interaction is supported.

Abstract Most graph visualization techniques focus on the structure of graphs and do not offer support for dealing with node attributes and edge labels. To enable users to detect relations and patterns in terms of data associated with nodes and edges, we present a technique where this data plays a more central role. Nodes and edges are clustered based on associated data. Via direct manipulation users can interactively inspect and query the graph. Questions that can be answered include, “which edge types are activated by specific node attributes?” and, “how and from where can I reach specific types of nodes?” To validate our approach we contrast it with current practice. We also provide several examples where our method was used to study transition graphs that model real‐world systems.

Abstract Color is widely used in data visualization to show data values. The proper selection of colors is critical to convey information correctly. In this paper, we present a technique for generating univariate lightness ordered palettes. These are specified via intuitive input parameters that are used define the appearance of the palette: number of colors, hue, lightness, saturation, contrast and hue range. The settings of the parameters are used to generate curves through CIELUV color space. This color space is used in order to correctly translate the requirements in terms of perceptual properties to a set of colors. The presented palette generation method enables users to specify palettes that have these perceptual properties, such as perceived order, equal perceived distance and equal importance. The technique has been integrated in Magna View, a system for multivariate data visualization.

We present a practical algorithm for computing robust multiscale curve and surface skeletons of 3D objects of genus zero. Based on a model that follows an advection principle, we assign to each point on the skeleton a part of the object surface, called the collapse. The size of the collapse is used as a uniform importance measure for the curve and surface skeleton, so that both can be simplified by imposing a single threshold on this intuitive measure. The simplified skeletons are connected by default, without special precautions, due to the monotonicity of the importance measure. The skeletons possess additional desirable properties: They are centered, robust to noise, hierarchical, and provide a natural skeleton-to-boundary mapping. We present a voxel-based algorithm that is straightforward to implement and simple to use. We illustrate our method on several realistic 3D objects.

The use of dynamic information to aid in software understanding is a common practice nowadays. One of the many approaches concerns the comprehension of execution traces. A major issue in this context is scalability: due to the vast amounts of information, it is a very difficult task to successfully find your way through such traces without getting lost. In this paper, we propose the use of a novel trace visualization method based on a massive sequence and circular bundle view, constructed with scalability in mind. By means of three usage scenarios that were conducted on three different software systems, we show how our approach, implemented in a tool called EXTRAVIS, is applicable to the areas of trace exploration, feature location, and feature comprehension.

Abstract Summary: The software package DNAVis offers a fast, interactive and real-time visualization of DNA sequences and their comparative genome annotations. DNAVis implements advanced methods of information visualization such as linked views, perspective walls and semantic zooming, in addition to the display of heterologous data in dot plot-like matrix views. Availability: The software is freely available at ; the source code is available upon request. Contact: [email protected] Supplementary information: Figures are available at Bioinformatics online; documentation and data are available at

Color is routinely used to visualize interval and ordinal data, while texture is not. For color, a variety of perceptually based models is available, which can be used to convey data via straightforward mapping. The dimensionality of texture is less well understood and there are almost no perceptually based and validated models available to generate textures on demand. We present a perceptually based texture synthesis model for isotropic textures. The model uses additive synthesis of band-limited noise in the spectral domain, comparable to the RGB (red, green, blue) model for color. Via user experiments, we have derived a three-dimensional model to control the amplitudes per band in a perceptually intuitive way, comparable to the HSV (hue, saturation, value) model for color. The three dimensions used are contrast, spatial frequency, and spectral purity. Besides a mapping to the amplitudes, we have derived a mapping to a perceptually equidistant space. We show how such textures can be combined with color; applications are presented.

ISUAL analytics is the science of analytical reasoning supportedbyhighlyinteractivevisualinterfaces.People use visual analytics tools and techniques to synthesize information; derive insight from massive, dynamic, and often conflicting data; detect the expected and discover the unexpected; provide timely, defensible, and understandable assessments; and communicate assessments effectively for action. The issues stimulating this body of research provide a grandchallengeinscience:turninginformationoverloadinto the opportunity of the decade. Visual analytics requires interdisciplinary science beyond traditional scientific and information visualization to include statistics, mathematics, knowledge representation, management and discovery technologies, cognitive and perceptual sciences, decision sciences, and more. An important research agenda to develop the next generation suite of visual analytics technologies is described in the book Illuminating the Path: The Research and Development Agenda for Visual Analytics, which is available at http:// nvac.pnl.gov/agenda.stm. The papers in this special section address a number of the issues described in the visual analytics research agenda. They are grouped into five major areas: multidimensional data, graphs and networks, communication network analysis, space and time, and fundamentals. The first two papers address issues of visual analysis of multidimensional data. The first paper, “High-Dimensional

We present a new approach for the visual analysis of state transition graphs. We deal with multivariate graphs where a number of attributes are associated with every node. Our method provides an interactive attribute-based clustering facility. Clustering results in metric, hierarchical and relational data, represented in a single visualization. To visualize hierarchically structured quantitative data, we introduce a novel technique: the bar tree. We combine this with a node-link diagram to visualize the hierarchy and an arc diagram to visualize relational data. Our method enables the user to gain significant insight into large state transition graphs containing tens of thousands of nodes. We illustrate the effectiveness of our approach by applying it to a real-world use case. The graph we consider models the behavior of an industrial wafer stepper and contains 55 043 nodes and 289 443 edges.

Business data is often presented using simple business graphics. These familiar visualizations are effective for providing overviews, but fall short for the presentation of large amounts of detailed information. Treemaps can provide such detail, but are often not easy to understand. We present how standard treemap algorithms can be adapted such that the results mimic familiar business graphics. Specifically, we present the use of different layout algorithms per level, a number of variations of the squarified algorithm, the use of variable borders, and the use of non-rectangular shapes. The combined use of these leads to histograms, pie charts and a variety of other styles.

The field of visualization is maturing. Many problems have been solved and new directions are sought. In order to make good choices, an understanding of the purpose and meaning of visualization is needed. In this paper, visualization is considered from multiple points of view. First, a technological viewpoint is adopted, where the value of visualization is measured based on effectiveness and efficiency. An economic model of visualization is presented and benefits and costs are established. Next, consequences and limitations of visualization are discussed (including the use of alternative methods, high initial costs, subjectiveness, and the role of interaction). Example uses of the model for the judgment of existing classes of methods are given to understand why they are or are not used in practice. However, such an economic view is too restrictive. Alternative views on visualization are presented and discussed: visualization as an art, visualization as design and, finally, visualization as a scientific discipline.

The genus of a knot or link can be defined via Seifert surfaces. A Seifert surface of a knot or link is an oriented surface whose boundary coincides with that knot or link. Schematic images of these surfaces are shown in every text book on knot theory, but from these it is hard to understand their shape and structure. In this paper, the visualization of such surfaces is discussed. A method is presented to produce different styles of surface for knots and links, starting from the so-called braid representation. Application of Seifert's algorithm leads to depictions that show the structure of the knot and the surface, while successive relaxation via a physically based model gives shapes that are natural and resemble the familiar representations of knots. Also, we present how to generate closed oriented surfaces in which the knot is embedded, such that the knot subdivides the surface into two parts. These closed surfaces provide a direct visualization of the genus of a knot. All methods have been integrated in a freely available tool, called SeifertView, which can be used for educational and presentation purposes.

Transition systems are graphs and it is natural to consider their visual representation as a graph visualization problem. They also pose an interesting multidimensional visualization challenge since every state may be considered as a point in n-space. We discuss a number of approaches toward projecting transition systems to the 2D plane by considering the dimensionality of the states. To assess these techniques, we consider their ability to assist us in answering questions that are difficult to address with conventional methods.

Fiber tracking is a standard approach for the visualization of the results of diffusion tensor imaging (DTI). If fibers are reconstructed and visualized individually through the complete white matter, the display gets easily cluttered making it difficult to get insight in the data. Various clustering techniques have been proposed to automatically obtain bundles that should represent anatomical structures, but it is unclear which clustering methods and parameter settings give the best results. We propose a framework to validate clustering methods for white-matter fibers. Clusters are compared with a manual classification which is used as a ground truth. For the quantitative evaluation of the methods, we developed a new measure to assess the difference between the ground truth and the clusterings. The measure was validated and calibrated by presenting different clusterings to physicians and asking them for their judgement. We found that the values of our new measure for different clusterings match well with the opinions of physicians. Using this framework, we have evaluated different clustering algorithms, including shared nearest neighbor clustering, which has not been used before for this purpose. We found that the use of hierarchical clustering using single-link and a fiber similarity measure based on the mean distance between fibers gave the best results.

The field of visualization is getting mature. Many problems have been solved, and new directions are sought for. In order to make good choices, an understanding of the purpose and meaning of visualization is needed. Especially, it would be nice if we could assess what a good visualization is. In this paper an attempt is made to determine the value of visualization. A technological viewpoint is adopted, where the value of visualization is measured based on effectiveness and efficiency. An economic model of visualization is presented, and benefits and costs are established. Next, consequences (brand limitations of visualization are discussed (including the use of alternative methods, high initial costs, subjective/less, and the role of interaction), as well as examples of the use of the model for the judgement of existing classes of methods and understanding why they are or are not used in practice. Furthermore, two alternative views on visualization are presented and discussed: viewing visualization as an art or as a scientific discipline. Implications and future directions are identified.

The genus of a knot or link can be defined via Seifert surfaces. A Seifert surface of a knot or link is an oriented surface whose boundary coincides with that, knot or link. Schematic images of these surfaces are shown in every text book on knot theory, but from these it is hard to understand their shape and structure. In this paper the visualization of such surfaces is discussed. A method is presented to produce different styles of surfaces for knots and links, starting from the so-called braid representation. Also, it is shown how closed oriented surfaces can be generated in which the knot is embedded, such that the knot subdivides the surface into two parts. These closed surfaces provide a direct visualization of the genus of a knot.

We present a side-by-side analysis of two recent image space approaches for the visualization of vector fields on surfaces. The two methods, Image Space Advection (ISA) and Image-Based Flow Visualization for Curved Surfaces (IBFVS) generate dense representations of time-dependent vector fields with high spatio-temporal correlation. While the 3D vector fields are associated with arbitrary surfaces represented by triangular meshes, the generation and advection of texture properties is confined to image space. Fast frame rates are achieved by exploiting frame-to-frame coherency and graphics hardware. In our comparison of ISA and IBFVS, we point out the strengths and weaknesses of each approach and give recommendations as to when and where they are best applied.

This issue contains extended versions of six of the strongest papers taken from the IEEE Visualization 2003 Conference. This year we have followed a slightly different procedure than in previous years. We intended to give prospective authors more time to prepare an extended version, while simultaneously keeping a short time between the conference and the release of this issue. Therefore, in June 2003, just after the selection of papers for the conference, we started to select the most interesting papers. We, paper cochairs of IEEE Visualization 2003, started from the opinions of the reviewers, followed by several rounds of rereading and discussion. The authors of six outstanding papers were invited to submit an extended and revised version of their original work. Each revised paper was thoroughly reviewed by several experts before it was recommended for acceptance. We are grateful to the reviewers for their thorough and quick reviews, and to David Ebert, Editor-in-Chief of TVCG, for his strong support, wise advise, and always astonishly quick responses. One of the most fascinating aspects of visualization is the width of the field. We selected the papers on quality, but we were delighted to find out that, in the end, our selection covers a wide variety of topics, approaches, and methods. Novel results on feature extraction and simplification are given, but also novel interpolation results; the handling of various standard types of data, including scalar volume data, terrain data, and vector volume data, is discussed, but also a new representation is presented; sophisticated mathematics, including group theory and Morse theory, is used, but also the capabilities of modern graphics hardware are exploited to the limit.

Large 2D information spaces, such as maps, images, or abstract visualizations, require views at various level of detail: close ups to inspect details, overviews to maintain (literally) an overview. Users often change their view during a session. Smooth animations enable the user to maintain an overview during interactive viewing and to understand the context of separate views. We present a generic model to handle smooth image viewing. The core of the model is a metric on the effect of simultaneous zooming and panning, based on an estimate of the perceived velocity. Using this metric, solutions for various problems are derived, such as the optimal animation between two views, automatic zooming, and the parametrization of arbitrary camera paths. Optimal is defined here as smooth and efficient. Solutions are based on the shortest paths of a virtual camera, given the metric. The model has two free parameters: animation speed and zoom/pan trade off. A user experiment to find good values for these is described. Finally, it is shown how the model can be extended to deal also with rotation and nonuniform scaling.

We present a hardware-accelerated method for visualizing 3D flow fields. The method is based on insertion, advection, and decay of dye. To this aim, we extend the texture-based IBFV technique presented by van Wijk (2001) for 2D flow visualization in two main directions. First, we decompose the 3D flow visualization problem in a series of 2D instances of the mentioned IBFV technique. This makes our method benefit from the hardware acceleration the original IBFV technique introduced. Secondly, we extend the concept of advected gray value (or color) noise by introducing opacity (or matter) noise. This allows us to produce sparse 3D noise pattern advections, thus address the occlusion problem inherent to 3D flow visualization. Overall, the presented method delivers interactively animated 3D flow, uses only standard OpenGL 1.1 calls and 2D textures, and is simple to understand and implement.

A new method for the synthesis of dense, vector-field aligned textures on curved surfaces is presented, called IBFVS. The method is based on image based flow visualization (IBFV). In IBFV two-dimensional animated textures are produced by defining each frame of a flow animation as a blend between a warped version of the previous image and a number of filtered white noise images. We produce flow aligned texture on arbitrary three-dimensional triangular meshes in the same spirit as the original method: texture is generated directly in image space. We show that IBFVS is efficient and effective. High performance (typically fifty frames or more per second) is achieved by exploiting graphics hardware. Also, IBFVS can easily be implemented and a variety of effects can be achieved. Applications are flow visualization and surface rendering. Specifically, we show how to visualize the wind field on the earth and how to render a dirty bronze bunny.

A new method for the visualization of two-dimensional fluid flow is presented. The method is based on the advection and decay of dye. These processes are simulated by defining each frame of a flow animation as a blend between a warped version of the previous image and a number of background images. For the latter a sequence of filtered white noise images is used: filtered in time and space to remove high frequency components. Because all steps are done using images, the method is named Image Based Flow Visualization ( IBFV ). With IBFV a wide variety of visualization techniques can be emulated. Flow can be visualized as moving textures with line integral convolution and spot noise. Arrow plots, streamlines, particles, and topological images can be generated by adding extra dye to the image. Unsteady flows, defined on arbitrary meshes, can be handled. IBFV achieves a high performance by using standard features of graphics hardware. Typically fifty frames per second are generated using standard graphics cards on PCs. Finally, IBFV is easy to understand, analyse, and implement.

A new method for the visualization of state transition systems is presented. Visual information is reduced by clustering nodes, forming a tree structure of related clusters. This structure is visualized in three dimensions with concepts from cone trees and emphasis on symmetry. A number of interactive options are provided as well, allowing the user to superimpose detail information on this tree structure. The resulting visualization enables the user to relate features in the visualization of the state transition graph to semantic concepts in the corresponding process and vice versa.

The visualization of scalar functions of two variables is a classic and ubiquitous application. We present a new method to visualize such data. The method is based on a nonlinear mapping of the function to a height field, followed by visualization as a shaded mountain landscape. The method is easy to implement and efficient, and leads to intriguing and insightful images: The visualization is enriched by adding ridges. Three types of applications are discussed: visualization of iso-levels, clusters (multivariate data visualization), and dense contours (flow visualization).

A new method for the simplification of flow fields is presented. It is based on continuous clustering. A well-known physical clustering model, the Cahn-Hilliard (1958) model, which describes phase separation, is modified to reflect the properties of the data to be visualized. Clusters are defined implicitly as connected components of the positivity set of a density function. An evolution equation for this function is obtained as a suitable gradient flow of an underlying anisotropic energy functional, where time serves as the scale parameter. The evolution is characterized by a successive coarsening of patterns, during which the underlying simulation data specifies preferable pattern boundaries. We introduce specific physical quantities in the simulation to control the shape, orientation and distribution of the clusters as a function of the underlying flow field. In addition, the model is expanded, involving elastic effects. In the early stages of the evolution, a shear-layer-type representation of the flow field can thereby be generated, whereas, for later stages, the distribution of clusters can be influenced. Furthermore, we incorporate upwind ideas to give the clusters an oriented drop-shaped appearance. We discuss the applicability of this new type of approach mainly for flow fields, where the cluster energy penalizes cross-streamline boundaries. However, the method also carries provisions for other fields as well. The clusters can be displayed directly as a flow texture. Alternatively, the clusters can be visualized by iconic representations, which are positioned by using a skeletonization algorithm.

A new method for the simplification of flow fields is presented. It is based on continuous clustering. A well-known physical clustering model, the Cahn Hilliard model (J. Cahn and J. Hilliard, 1958), which describes phase separation, is modified to reflect the properties of the data to be visualized. Clusters are defined implicitly as connected components of the positivity set of a density function. An evolution equation for this function is obtained as a suitable gradient flow of an underlying anisotropic energy functional. Here, time serves as the scale parameter. The evolution is characterized by a successive coarsening of patterns: the actual clustering, and meanwhile the underlying simulation data specifies preferable pattern boundaries. The authors discuss the applicability of this new type of approach mainly for flow fields, where the cluster energy penalizes cross streamline boundaries, but the method also carries provisions in other fields as well. The clusters are visualized via iconic representations. A skeletonization algorithm is used to find suitable positions for the icons.

Vector field visualization remains a difficult task. Many local and global visualization methods for vector fields such as flow data exist, but they usually require extensive user experience on setting the visualization parameters in order to produce images communicating the desired insight. We present a visualization method that produces simplified but suggestive images of the vector field automatically, based on a hierarchical clustering of the input data. The resulting clusters are then visualized with straight or curved arrow icons. The presented method has a few parameters with which users can produce various simplified vector field visualizations that communicate different insights on the vector data.

Rendering objects transparently gives additional insight in complex and overlapping structures. However, traditional techniques for the rendering of transparent objects such as alpha blending are not very well suited for the rendering of multiple transparent objects in dynamic scenes. Screen door transparency is a technique to render transparent objects in a simple and efficient way: no sorting is required and intersecting polygons can be handled without further preprocessing. With this technique, polygons are rendered through a mask: only where the mask is present, pixels are set. However, artifacts such as incorrect opacities and distracting patterns can easily occur if the masks are not carefully designed. The requirements on the masks are considered. Next, three algorithms are presented for the generation of pixel masks. One algorithm is designed for the creation of small (e.g. 4/spl times/4) masks. The other two algorithms can be used for the creation of larger masks (e.g. 32/spl times/32). For each of these algorithms, results are presented and discussed.

We present a system where visualization and the control of the simulation are integrated to facilitate interactive exploration and modeling of large data sets. The system was developed to estimate properties of the atmosphere of Venus from comparison between measured and simulated data. Reuse of results, distributed computing, and multiple views on the data were the major ingredients to create an effective environment.

Spot noise is a technique for texture synthesis, which is very useful for vector field visualization. This paper describes improvements and extensions of the basic principle of spot noise. First, better visualization of highly curved vector fields with spot noise is achieved, by adapting the shape of the spots to the local velocity field. Second, filtering of spots is proposed to eliminate undesired low frequency components from the spot noise texture. Third, methods are described to utilize graphics hardware to generate the texture, and to produce variable viewpoint animations of spot noise on surfaces. Fourth, the synthesis of spot noise on grids with highly irregular cell sizes is described.

Computational steering is the ultimate goal of interactive simulation: researchers change parameters of their simulation and immediately receive feedback on the effect. We present a general and flexible graphics tool that is part of an environment for computational steering developed at CWI. It enables the researcher to interactively develop his own interface with the simulation. This interface is constructed with 3D parametrized geometric objects. The properties of the objects are parametrized to output data and input parameters of the simulation. The objects visualize the output of the simulation, while the researcher can steer the simulation by direct manipulation of the objects. Several applications of 3D computational steering are presented.

A probe for the interactive visualization of flow fields is presented. The probe can be used to visualize many characteristics of the flow in detail for a small region in the data set. The velocity and the local change of velocity (the velocity gradient tensor) are visualized by a set of geometric primitives. To this end, the velocity gradient tensor is transformed to a local coordinate frame, and decomposed into components parallel with and perpendicular to the flow. These components are visualized as geometric objects with an intuitively meaningful interpretation. An implementation is presented which shows that this probe is a useful tool for flow visualization.>

Streamlines and stream surfaces are well known techniques for the visualization of fluid flow. For steady velocity fields, a streamline is the trace of a particle, and a stream surface is the trace of a curve. Here a new method is presented for the construction of stream surfaces. The central concept is the representation of a stream surface as an implicit surface f (x) = C. After the initial calculation of f a family of stream surfaces can be generated efficiently by varying C. The shapes of the originating curves are defined by the value of f at the boundary. Two techniques are presented for the calculation of f: one based on solving the convection equation, the other on backward tracing of the trajectories of grid points. The flow around objects is discussed separately. With this method irregular topologies of the originating curves and of the stream surfaces can be handled easily. Further, it can also be used for other visualization techniques, such as time surfaces and stream volumes. Finally, an effective method for the automatic placement of originating curves is presented.>

HyperSlice is a new method for the visualization of scalar functions of many variables. With this method the multi-dimensional function is presented in a simple and easy to understand way in which all dimensions are treated identically. The central concept is the representation of a multi-dimensional function as a matrix of orthogonal two-dimensional slices. These two-dimensional slices lend themselves very well to interaction via direct manipulation, due to a one to one relation between screen space and variable space. Several interaction techniques, for navigation, the location of maxima, and the use of user-defined paths, are presented.>

Surface-particles are very small facets, modeled as points with a normal. They can be used to visualize flow in several ways by variation of the properties of the particle sources. A method is presented for the rendering of surface-particles. This method includes an improved shading model, the use of Gaussian filters for the prevention of spatial and temporal artifacts, an efficient scan-conversion algorithm, the handling of occlusion and the simultaneous rendering of geometric objects and surface-particles. The synthesis of images with limited depth of field is described, which literally allows the scientist to focus on areas of interest.>

The use of stochastic texture for the visualization of scalar and vector fields over surfaces is discussed. Current techniques for texture synthesis are not suitable, because they do not provide local control, and are not suited for the design of textures. A new technique, spot noise, is presented that does provide these features. Spot noise is synthesized by addition of randomly weighted and positioned spots. Local control of the texture is realized by variation of the spot. The spot is a useful primitive for texture design, because, in general, the relations between features of the spot and features of the texture are straightforward. Various examples and applications are shown. Spot noise lends itself well for the synthesis of texture over curved surfaces, and is therefore an alternative to solid texturing. The relations of spot noise with a variety of other techniques, such as radom faults, filtering, sparse convolution, and particle systems, are discussed. It appears that spot noise provides a new perspective on those techniques.

Most techniques for the visualization of fluid dynamics are based on vector graphics. Streamlines or arrows are calculated as vectors, which are converted to pixels for display on raster devices. The techniques described here show that it can be advantageous to use a raster graphics approach to the visualization of fluid dynamics. The properties of raster devices can be exploited better, while also anti-aliasing can be realized in a simple way. Three techniques for visualization of two-dimensional flow are described: for the calculation and display of streamlines; an alternative technique based on particle systems for capturing flow in a still image; and for the creation of animations using particles.

Abstract Now that raster graphics has become a common tool in CAD/CAM systems, animation and many other applications, the need has arisen for user interfaces to graphics packages for these new applications. On the basis of a typology of the users, an analysis is presented of some of the choices which have to be made in developing these interfaces.

The crucial step in a program based on ray tracing is the calculation of the intersection of a line with an object.In this paper, algorithms are presented for performing this calculation for objects defined by sweeping a planar cubic spline through space.Translational, rotational, and conic sweeping are treated.Besides solutions for the exact calculation, rectangle tests for improving efficiency are given.Possible extensions and improvements are discussed.

The long processing times inherent in the use of raster graphics techniques for the shading of complex scenes, complicate the design and debugging of a picture. The efficiency of the picture-generating process can, however, be improved by adding several interactive facilities to the system. In particular, fast previewing techniques are valuable for checking the modeling and for getting a good impression in an early stage of the picture-definition process. A discussion of raster graphics system design is followed by examples of fast pre-viewing techniques for the shading and texture mapping of surface models and for the shading of solid models using a raycasting technique.