Generative models are becoming a transformative technology for the creation and editing of images. However, it remains challenging to harness these models for precise image manipulation. These challenges often manifest as inconsistency in the editing process, where both the type and amount of semantic change, depend on the image being manipulated. Moreover, there exist many methods for computing image manipulations, whose development is hindered by the matter of inconsistency. This paper aims to address these challenges by improving how we evaluate, compare, and explore the space of manipulations offered by a generative model. We present Concept Lens, a visual interface that is designed to aid users in understanding semantic concepts carried in image manipulations, and how these manipulations vary over generated images. Given the large space of possible images produced by a generative model, Concept Lens is designed to support the exploration of both generated images, and their manipulations, at multiple levels of detail. To this end, the layout of Concept Lens is informed by two hierarchies: a hierarchical organization of (1) original images, grouped by their similarities, and (2) image manipulations, where manipulations that induce similar changes are grouped together. This layout allows one to discover the types of images that consistently respond to a group of manipulations, and vice versa, manipulations that consistently respond to a group of codes. We show the benefits of this design across multiple use cases, specifically, studying the quality of manipulations for a single method, and offering a means of comparing different methods.

Deep learning approaches have provided state-of-the-art performance in many applications by relying on large and overparameterized neural networks. However, such networks are very brittle and are difficult to deploy on resource-limited platforms. Model pruning, i.e., reducing the size of the network, is a widely adopted strategy that can lead to a more robust and compact model. Many heuristics exist for model pruning, but our understanding of the pruning process remains limited due to the black-box nature of a neural network model. Empirical studies show that some heuristics improve performance whereas others can make models more brittle. This work aims to shed light on how different pruning methods alter the network's internal feature representation and the corresponding impact on model performance. To facilitate a comprehensive comparison and characterization of the high-dimensional model feature space, we introduce a visual geometric analysis of feature representations. We evaluated a set of critical geometric concepts decomposed from the commonly adopted classification loss and used them to design a visualization system to compare and highlight the impact of pruning on model performance and feature representation. The proposed tool provides an environment for an in-depth comparison of pruning methods and a comprehensive understanding of how the model responds to common data corruption. By leveraging the proposed visualization, machine learning researchers can reveal the similarities between pruning methods and redundancy in robustness evaluation benchmarks, obtain geometric insights about the differences between pruned models that achieve superior robustness performance, and identify samples that are robust or fragile to model pruning and common data corruption.

Transfer function design is crucial in volume rendering, as it directly influences the visual representation and interpretation of volumetric data. However, creating effective transfer functions that align with users’ visual objectives is often challenging due to the complex parameter space and the semantic gap between transfer function values and features of interest within the volume. In this work, we propose a novel approach that leverages recent advancements in language-vision models to bridge this semantic gap. By employing a fully differentiable rendering pipeline and an image-based loss function guided by language descriptions, our method generates transfer functions that yield volume-rendered images closely matching the user’s intent. We demonstrate the effectiveness of our approach in creating meaningful transfer functions from simple descriptions, empowering users to intuitively express their desired visual outcomes with minimal effort. This advancement streamlines the transfer function design process and makes volume rendering more accessible to a wider range of users.

Abstract We present concept‐aligned neurons, or CAN, a visualization design for comparing deep neural networks. The goal of CAN is to support users in understanding the similarities and differences between neural networks, with an emphasis on comparing neuron functionality across different models. To make this comparison intuitive, CAN uses concept‐based representations of neurons to visually align models in an interpretable manner. A key feature of CAN is the hierarchical organization of concepts, which permits users to relate sets of neurons at different levels of detail. CAN's visualization is designed to help compare the semantic coverage of neurons, as well as assess the distinctiveness, redundancy, and multi‐semantic alignment of neurons or groups of neurons, all at different concept granularity. We demonstrate the generality and effectiveness of CAN by comparing models trained on different datasets, neural networks with different architectures, and models trained for different objectives, e.g. adversarial robustness, and robustness to out‐of‐distribution data.

Abstract With recent advances in multi‐modal foundation models, the previously text‐only large language models (LLM) have evolved to incorporate visual input, opening up unprecedented opportunities for various applications in visualization. Compared to existing work on LLM‐based visualization works that generate and control visualization with textual input and output only, the proposed approach explores the utilization of the visual processing ability of multi‐modal LLMs to develop Autonomous Visualization Agents (AVAs) that can evaluate the generated visualization and iterate on the result to accomplish user‐defined objectives defined through natural language. We propose the first framework for the design of AVAs and present several usage scenarios intended to demonstrate the general applicability of the proposed paradigm. Our preliminary exploration and proof‐of‐concept agents suggest that this approach can be widely applicable whenever the choices of appropriate visualization parameters require the interpretation of previous visual output. Our study indicates that AVAs represent a general paradigm for designing intelligent visualization systems that can achieve high‐level visualization goals, which pave the way for developing expert‐level visualization agents in the future.

High-performance computing (HPC) systems play a critical role in facilitating scientific discoveries. Their scale and complexity (e.g., the number of computational units and software stack) continue to grow as new systems are expected to process increasingly more data and reduce computing time. However, with more processing elements, the probability that these systems will experience a random bit-flip error that corrupts a program's output also increases, which is often recognized as silent data corruption. Analyzing the resiliency of HPC applications in extreme-scale computing to silent data corruption is crucial but difficult. An HPC application often contains a large number of computation units that need to be tested, and error propagation caused by error corruption is complex and difficult to interpret. To accommodate this challenge, we propose an interactive visualization system that helps HPC researchers understand the resiliency of HPC applications and compare their error propagation. Our system models an application's error propagation to study a program's resiliency by constructing and visualizing its fault tolerance boundary. Coordinating with multiple interactive designs, our system enables domain experts to efficiently explore the complicated spatial and temporal correlation between error propagations. At the end, the system integrated a nonmonotonic error propagation analysis with an adjustable graph propagation visualization to help domain experts examine the details of error propagation and answer such questions as why an error is mitigated or amplified by program execution.

Generative Adversarial Networks (GANs) are notoriously difficult to train especially for complex distributions and with limited data. This has driven the need for tools to audit trained networks in human intelligible format, for example, to identify biases or ensure fairness. Existing GAN audit tools are restricted to coarse-grained, modeldata comparisons based on summary statistics such as FID or recall. In this paper, we propose an alternative approach that compares a newly developed GAN against a prior baseline. To this end, we introduce Cross-GAN Auditing (xGA) that, given an established “reference” GAN and a newly proposed “client” GAN, jointly identifies intelligible attributes that are either common across both GANs, novel to the client GAN, or missing from the client GAN. This provides both users and model developers an intuitive assessment of similarity and differences between GANs. We introduce novel metrics to evaluate attribute-based GAN auditing approaches and use these metrics to demonstrate quantitatively that xGA outperforms baseline approaches. We also include qualitative results that illustrate the common, novel and missing attributes identified by xGA from GANs trained on a variety of image datasets1

Rich semantics exist in latent spaces inferred using deep generative models. The ability to extract and interpret them is not only essential for understanding the underlying factors of variation in the data distribution, but also crucial for con-trolled image generation. Several methods have been proposed to identify semantically meaningful linear directions, either through existing annotations, or relying on identifying directions of large variation that arise from the data representation of the network. In this paper, we identify a new criterion, representation sparsity, that allows us to produce extremely efficient yet diverse semantic directions in GAN (generative adversarial network) latent spaces. The observation also reveals a potential deeper connection between representation sparsity and semantics in deep neural networks that worth further exploration.

Abstract Generative adversarial networks (GAN) have witnessed tremendous growth in recent years, demonstrating wide applicability in many domains. However, GANs remain notoriously difficult for people to interpret, particularly for modern GANs capable of generating photo‐realistic imagery. In this work we contribute a visual analytics approach for GAN interpretability, where we focus on the analysis and visualization of GAN disentanglement. Disentanglement is concerned with the ability to control content produced by a GAN along a small number of distinct, yet semantic, factors of variation. The goal of our approach is to shed insight on GAN disentanglement, above and beyond coarse summaries, instead permitting a deeper analysis of the data distribution modeled by a GAN. Our visualization allows one to assess a single factor of variation in terms of groupings and trends in the data distribution, where our analysis seeks to relate the learned representation space of GANs with attribute‐based semantic scoring of images produced by GANs. Through use‐cases, we show that our visualization is effective in assessing disentanglement, allowing one to quickly recognize a factor of variation and its overall quality. In addition, we show how our approach can highlight potential dataset biases learned by GANs.

The trend of rapid technology scaling is expected to make the hardware of high-performance computing (HPC) systems more susceptible to computational errors due to random bit flips. Some bit flips may cause a program to crash or have a minimal effect on the output, but others may lead to silent data corruption (SDC), i.e., undetected yet significant output errors. Classical fault injection analysis methods employ uniform sampling of random bit flips during program execution to derive a statistical resiliency profile. However, summarizing such fault injection result with sufficient detail is difficult, and understanding the behavior of the fault-corrupted program is still a challenge. In this article, we introduce SpotSDC, a visualization system to facilitate the analysis of a program's resilience to SDC. SpotSDC provides multiple perspectives at various levels of detail of the impact on the output relative to where in the source code the flipped bit occurs, which bit is flipped, and when during the execution it happens. SpotSDC also enables users to study the code protection and provide new insights to understand the behavior of a fault-injected program. Based on lessons learned, we demonstrate how what we found can improve the fault injection campaign method.

With the rapid adoption of machine learning techniques for large-scale applications in science and engineering comes the convergence of two grand challenges in visualization. First, the utilization of black box models (e.g., deep neural networks) calls for advanced techniques in exploring and interpreting model behaviors. Second, the rapid growth in computing has produced enormous datasets that require techniques that can handle millions or more samples. Although some solutions to these interpretability challenges have been proposed, they typically do not scale beyond thousands of samples, nor do they provide the high-level intuition scientists are looking for. Here, we present the first scalable solution to explore and analyze high-dimensional functions often encountered in the scientific data analysis pipeline. By combining a new streaming neighborhood graph construction, the corresponding topology computation, and a novel data aggregation scheme, namely topology aware datacubes, we enable interactive exploration of both the topological and the geometric aspect of high-dimensional data. Following two use cases from high-energy-density (HED) physics and computational biology, we demonstrate how these capabilities have led to crucial new insights in both applications.

Training deep neural networks on large scientific data is a challenging task that requires enormous compute power, especially if no pre-trained models exist to initialize the process. We present a novel tournament method to train traditional as well as generative adversarial networks built on LBANN, a scalable deep learning framework optimized for HPC systems. LBANN combines multiple levels of parallelism and exploits some of the worlds largest supercomputers.We demonstrate our framework by creating a complex predictive model based on multi-variate data from high-energy-density physics containing hundreds of millions of images and hundreds of millions of scalar values derived from tens of millions of simulations of inertial confinement fusion. Our approach combines an HPC workflow and extends LBANN with optimized data ingestion and the new tournament-style training algorithm to produce a scalable neural network architecture using a CORAL-class supercomputer. Experimental results show that 64 trainers (1024 GPUs) achieve a speedup of 70.2× over a single trainer (16 GPUs) baseline, and an effective 109% parallel efficiency.

With the recent advances in deep learning, neural network models have obtained state-of-the-art performances for many linguistic tasks in natural language processing. However, this rapid progress also brings enormous challenges. The opaque nature of a neural network model leads to hard-to-debug-systems and difficult-to-interpret mechanisms. Here, we introduce a visualization system that, through a tight yet flexible integration between visualization elements and the underlying model, allows a user to interrogate the model by perturbing the input, internal state, and prediction while observing changes in other parts of the pipeline. We use the natural language inference problem as an example to illustrate how a perturbation-driven paradigm can help domain experts assess the potential limitation of a model, probe its inner states, and interpret and form hypotheses about fundamental model mechanisms such as attention.

Abstract Two‐dimensional embeddings remain the dominant approach to visualize high dimensional data. The choice of embeddings ranges from highly non‐linear ones, which can capture complex relationships but are difficult to interpret quantitatively, to axis‐aligned projections, which are easy to interpret but are limited to bivariate relationships. Linear project can be considered as a compromise between complexity and interpretability, as they allow explicit axes labels, yet provide significantly more degrees of freedom compared to axis‐aligned projections. Nevertheless, interpreting the axes directions, which are often linear combinations of many non‐trivial components, remains difficult. To address this problem we introduce a structure aware decomposition of (multiple) linear projections into sparse sets of axis‐aligned projections, which jointly capture all information of the original linear ones. In particular, we use tools from Dempster‐Shafer theory to formally define how relevant a given axis‐aligned project is to explain the neighborhood relations displayed in some linear projection. Furthermore, we introduce a new approach to discover a diverse set of high quality linear projections and show that in practice the information of k linear projections is often jointly encoded in ∼ k axis‐aligned plots. We have integrated these ideas into an interactive visualization system that allows users to jointly browse both linear projections and their axis‐aligned representatives. Using a number of case studies we show how the resulting plots lead to more intuitive visualizations and new insights.

Constructing distributed representations for words through neural language models and using the resulting vector spaces for analysis has become a crucial component of natural language processing (NLP). However, despite their widespread application, little is known about the structure and properties of these spaces. To gain insights into the relationship between words, the NLP community has begun to adapt high-dimensional visualization techniques. In particular, researchers commonly use t-distributed stochastic neighbor embeddings (t-SNE) and principal component analysis (PCA) to create two-dimensional embeddings for assessing the overall structure and exploring linear relationships (e.g., word analogies), respectively. Unfortunately, these techniques often produce mediocre or even misleading results and cannot address domain-specific visualization challenges that are crucial for understanding semantic relationships in word embeddings. Here, we introduce new embedding techniques for visualizing semantic and syntactic analogies, and the corresponding tests to determine whether the resulting views capture salient structures. Additionally, we introduce two novel views for a comprehensive study of analogy relationships. Finally, we augment t-SNE embeddings to convey uncertainty information in order to allow a reliable interpretation. Combined, the different views address a number of domain-specific tasks difficult to solve with existing tools.

Massive simulations and arrays of sensing devices, in combination with increasing computing resources, have generated large, complex, high-dimensional datasets used to study phenomena across numerous fields of study. Visualization plays an important role in exploring such datasets. We provide a comprehensive survey of advances in high-dimensional data visualization that focuses on the past decade. We aim at providing guidance for data practitioners to navigate through a modular view of the recent advances, inspiring the creation of new visualizations along the enriched visualization pipeline, and identifying future opportunities for visualization research.

Abstract Linear projections are one of the most common approaches to visualize high‐dimensional data. Since the space of possible projections is large, existing systems usually select a small set of interesting projections by ranking a large set of candidate projections based on a chosen quality measure. However, while highly ranked projections can be informative, some lower ranked ones could offer important complementary information. Therefore, selection based on ranking may miss projections that are important to provide a global picture of the data. The proposed work fills this gap by presenting the Grassmannian Atlas, a framework that captures the global structures of quality measures in the space of all projections, which enables a systematic exploration of many complementary projections and provides new insights into the properties of existing quality measures.

Abstract We introduce a novel interactive framework for visualizing and exploring high‐dimensional datasets based on subspace analysis and dynamic projections. We assume the high‐dimensional dataset can be represented by a mixture of low‐dimensional linear subspaces with mixed dimensions, and provide a method to reliably estimate the intrinsic dimension and linear basis of each subspace extracted from the subspace clustering. Subsequently, we use these bases to define unique 2D linear projections as viewpoints from which to visualize the data. To understand the relationships among the different projections and to discover hidden patterns, we connect these projections through dynamic projections that create smooth animated transitions between pairs of projections. We introduce the view transition graph, which provides flexible navigation among these projections to facilitate an intuitive exploration. Finally, we provide detailed comparisons with related systems, and use real‐world examples to demonstrate the novelty and usability of our proposed framework.

Abstract Dimension reduction techniques are essential for feature selection and feature extraction of complex high‐dimensional data. These techniques, which construct low‐dimensional representations of data, are typically geometrically motivated, computationally efficient and approximately preserve certain structural properties of the data. However, they are often used as black box solutions in data exploration and their results can be difficult to interpret. To assess the quality of these results, quality measures, such as co‐ranking [ LV09 ], have been proposed to quantify structural distortions that occur between high‐dimensional and low‐dimensional data representations. Such measures could be evaluated and visualized point‐wise to further highlight erroneous regions [ MLGH13 ]. In this work, we provide an interactive visualization framework for exploring high‐dimensional data via its two‐dimensional embeddings obtained from dimension reduction, using a rich set of user interactions. We ask the following question: what new insights do we obtain regarding the structure of the data, with interactive manipulations of its embeddings in the visual space? We augment the two‐dimensional embeddings with structural abstractions obtained from hierarchical clusterings, to help users navigate and manipulate subsets of the data. We use point‐wise distortion measures to highlight interesting regions in the domain, and further to guide our selection of the appropriate level of clusterings that are aligned with the regions of interest. Under the static setting, point‐wise distortions indicate the level of structural uncertainty within the embeddings. Under the dynamic setting, on‐the‐fly updates of point‐wise distortions due to data movement and data deletion reflect structural relations among different parts of the data, which may lead to new and valuable insights.

This paper evaluates features of graph coloring algorithms implemented on graphics processing units (GPUs), comparing coloring heuristics and thread decompositions. As compared to prior work on graph coloring for other parallel architectures, we find that the large number of cores and relatively high global memory bandwidth of a GPU lead to different strategies for the parallel implementation. Specifically, we find that a simple uniform block partitioning is very effective on GPUs and our parallel coloring heuristics lead to the same or fewer colors than prior approaches for distributed-memory cluster architecture. Our algorithm resolves many coloring conflicts across partitioned blocks on the GPU by iterating through the coloring process, before returning to the CPU to resolve remaining conflicts. With this approach we get as few color (if not fewer) than the best sequential graph coloring algorithm and performance is close to the fastest sequential graph coloring algorithms which have poor color quality.

We present a new framework for feature-based statistical analysis of large-scale scientific data and demonstrate its effectiveness by analyzing features from Direct Numerical Simulations (DNS) of turbulent combustion. Turbulent flows are ubiquitous and account for transport and mixing processes in combustion, astrophysics, fusion, and climate modeling among other disciplines. They are also characterized by coherent structure or organized motion, i.e. nonlocal entities whose geometrical features can directly impact molecular mixing and reactive processes. While traditional multi-point statistics provide correlative information, they lack nonlocal structural information, and hence, fail to provide mechanistic causality information between organized fluid motion and mixing and reactive processes. Hence, it is of great interest to capture and track flow features and their statistics together with their correlation with relevant scalar quantities, e.g. temperature or species concentrations. In our approach we encode the set of all possible flow features by pre-computing merge trees augmented with attributes, such as statistical moments of various scalar fields, e.g. temperature, as well as length-scales computed via spectral analysis. The computation is performed in an efficient streaming manner in a pre-processing step and results in a collection of meta-data that is orders of magnitude smaller than the original simulation data. This meta-data is sufficient to support a fully flexible and interactive analysis of the features, allowing for arbitrary thresholds, providing per-feature statistics, and creating various global diagnostics such as Cumulative Density Functions (CDFs), histograms, or time-series. We combine the analysis with a rendering of the features in a linked-view browser that enables scientists to interactively explore, visualize, and analyze the equivalent of one terabyte of simulation data. We highlight the utility of this new framework for combustion science; however, it is applicable to many other science domains.