Contemporary materials science research is heavily conducted in silico, involving massive simulations of the atomic-scale evolution of materials. Cataloging basic patterns in the atomic displacements is key to understanding and predicting the evolution of physical properties. However, the combinatorial complexity of the space of possible transitions coupled with the overwhelming amount of data being produced by high-throughput simulations make such an analysis extremely challenging and time-consuming for domain experts. The development of visual analytics systems that facilitate the exploration of simulation data is an active field of research. While these systems excel in identifying temporal regions of interest, they treat each timestep of a simulation as an independent event without considering the behavior of the atomic displacements between timesteps. We address this gap by introducing LAMDA, a visual analytics system that allows domain experts to quickly and systematically explore state-to-state transitions. In LAMDA, transitions are hierarchically categorized, providing a basis for cataloging displacement behavior, as well as enabling the analysis of simulations at different resolutions, ranging from very broad qualitative classes of transitions to very narrow definitions of unit processes. LAMDA supports navigating the hierarchy of transitions, enabling scientists to visualize the commonalities between different transitions in each class in terms of invariant features characterizing local atomic environments, and LAMDA simplifies the analysis by capturing user inputs through annotations. We evaluate our system through a case study and report on findings from our domain experts.

The proliferation of misinformation in journalism, often stemming from flawed reasoning and logical fallacies, poses significant challenges to public understanding and trust in news media. Traditional fact-checking methods, while valuable, are insufficient for detecting the subtle logical inconsistencies that can mislead readers within seemingly factual content. To address this gap, we introduce Skeptik, a hybrid framework that integrates Large Language Models (LLMs) with heuristic approaches to analyze and annotate potential logical fallacies and reasoning errors in online news articles. Operating as a web browser extension, Skeptik automatically highlights sentences that may contain logical fallacies, provides detailed explanations, and offers multi-layered interventions to help readers critically assess the information presented. The system is designed to be extensible, accommodating a wide range of fallacy types and adapting to evolving misinformation tactics. Through comprehensive case studies, quantitative analyses, usability experiments, and expert evaluations, we demonstrate the effectiveness of Skeptik in enhancing readers' critical examination of news content and promoting media literacy. Our contributions include the development of an expandable classification system for logical fallacies, the innovative integration of LLMs for real-time analysis and annotation, and the creation of an interactive user interface that fosters user engagement and close reading. By emphasizing the logical integrity of textual content rather than relying solely on factual accuracy, Skeptik offers a comprehensive solution to combat potential misinformation in journalism. Ultimately, our framework aims to improve critical reading and protect the public from deceptive information online and enhance the overall credibility of news media.

The interdependencies of food, energy, and water (FEW) systems create a nexus opportunity to explore the strengths and vulnerabilities of individual and cross-sector interactions within FEW systems. However, the variables quantifying nexus interactions are hard to observe, which hinders the cross-sector analysis. To overcome such challenges, we present FEWSim, a visual analytics framework designed to support domain experts in exploring and interpreting simulation results from a coupled FEW model. FEWSim employs a three-layer asynchronous architecture: the model layer integrates food, energy, and water models to simulate the FEW nexus; the middleware layer manages scenario configuration and execution; and the visualization layer provides interactive visual exploration of simulated time-series results across FEW sectors. The visualization layer further facilitates the exploration across multiple scenarios and evaluates scenario differences in performance using sustainability indices of the FEW nexus. We demonstrate the utility of FEWSim through a case study for the Phoenix Active Management Area (AMA) in Arizona.

As Large Language Models (LLMs) become more popular within the visualization community, researchers increasingly leverage them for diverse visualization tasks such as design guideline suggestions and visualization evaluation. However, in order for LLMs to act as trustworthy and fair evaluators, we argue that LLMs would need to possess visualization literacy, be capable of following user instructions and uphold graphical integrity. We test the latest versions of the most prominent LLMs, specifically Anthropic's Claude (Opus 4.5), OpenAI's Generative Pretrained Transformers (GPT 5.2 Pro), and Google's Gemini (Gemini 3 Flash) on these features and find that while these models now possess visualization literacy, they still struggle with other features necessary for instruction following and graphical integrity. Using a modified Visualization Literacy Assessment Test (VLAT), our findings show that these recent LLMs have achieved greater than human-levels of visualization literacy in contrast to prior research. In order to test the models' abilities to follow instructions, we used few-shot and chain-of-thought prompting as proxies for instruction following tasks on evaluating visualization literacy and find that these specialized prompting techniques are becoming obsolete with respect to improving visualization literacy. Additionally, we experiment with the inherent ability of LLMs to evaluate misleading visualizations to test the models' abilities for upholding graphical integrity and find that without specialized or leading prompting techniques, the models struggle with being able to accurately identify whether a visualization is misleading or not. Our results further break down the performance of each model on these tasks, but the culmination of our findings force us to reconsider the current effectiveness of LLMs as visualization evaluators.

The rational development of transition-metal catalysts, even when guided by theory and mechanistic knowledge, involves significant trial and error. Although machine learning offers the potential to accelerate catalyst discovery and optimization, accurately modeling the complex structures of catalysts and the multistep mechanisms by which they react remains challenging, given the limited sets of data available. Olefin hydroformylation is a quintessential example of this challenge: its catalytic cycle involves many, often reversible, steps, and decades of study have not yielded reliable structure-selectivity relationships. We report Libra-ML, a 3D structure-based deep learning approach for predicting experimental outcomes of transition-metal-catalyzed reactions. To demonstrate the ability of Libra-ML to model the outcomes of complex catalytic reactions, we predicted the regioselectivity of hydroformylation with terminal olefins catalyzed by rhodium complexes. Comparisons to existing methods demonstrate the strong performance of Libra-ML and illustrate the importance of encoding 3D structures to predict experimental outcomes with molecular catalysts.

Loss landscapes are a powerful tool for understanding neural network optimization and generalization, yet traditional low-dimensional analyses often miss complex topological features. We present Landscaper, an open-source Python package for arbitrary-dimensional loss landscape analysis. Landscaper combines Hessian-based subspace construction with topological data analysis to reveal geometric structures such as basin hierarchy and connectivity. A key component is the Saddle-Minimum Average Distance (SMAD) for quantifying landscape smoothness. We demonstrate Landscaper's effectiveness across various architectures and tasks, including those involving pre-trained language models, showing that SMAD captures training transitions, such as landscape simplification, that conventional metrics miss. We also illustrate Landscaper's performance in challenging chemical property prediction tasks, where SMAD can serve as a metric for out-of-distribution generalization, offering valuable insights for model diagnostics and architecture design in data-scarce scientific machine learning scenarios.

Chemists and materials scientists increasingly use machine learning models, such as graph neural networks (GNNs), to predict properties of molecules and the outcomes of their reactions. Beyond predictive performance, understanding how these models organize chemical information internally in their latent spaces, i.e., the embeddings of the molecules, is critical. Analyzing latent spaces helps diagnose model behavior and assess whether the learned embeddings are organized in ways that reflect meaningful chemical relationships. Unfortunately, existing methods provide limited support for analyzing latent spaces across layers and across different model states (e.g., training epochs, model configurations, and input data), making it difficult to understand how these latent spaces evolve throughout a model or relate to chemical concepts. We present LatentFlow, a visual analytics system developed in collaboration with a domain expert for analyzing latent spaces in molecular GNNs. LatentFlow groups embeddings into clusters and supports exploration of latent spaces by tracking how these clusters change across layers and model states using a modified Sankey diagram. To support interpretation, LatentFlow links these clusters to representative molecules and their shared substructures, and it allows scientists to introduce their own domain knowledge and compare it with the patterns found in the latent spaces. We evaluate LatentFlow through two case studies. The results show that LatentFlow helps scientists understand how latent spaces evolve, identify meaningful molecular patterns, and better interpret model behavior.

Thematic maps are widely used to communicate spatial patterns to non-expert audiences. Although uncertainty is inherent in thematic map data, it is rarely visualized, raising questions about how its inclusion affects trust. Prior work offers mixed perspectives: some argue that uncertainty fosters trust through transparency, while others suggest it may reduce trust by introducing confusion. Yet few empirical studies explicitly measure trust in thematic maps. We conducted a between-subjects experiment (N=161) to evaluate how visualizing uncertainty at varying levels (low, medium, high) influences trust. We find that uncertainty visualization generally reduces trust, with greater reductions observed as uncertainty levels increase. However, maps dominated by low uncertainty do not significantly differ in trust from those with no uncertainty. Moreover, while uncertainty visualization tends to make readers question the accuracy of the data, it appears to have a weaker influence on perceptions of the mapmaker's integrity.

In this article, we assess the visualization literacy of two prominent Large Language Models (LLMs): OpenAI's Generative Pretrained Transformers (GPT), the backend of ChatGPT, and Google's Gemini, previously known as Bard, to establish benchmarks for assessing their visualization capabilities. While LLMs have shown promise in generating chart descriptions, captions, and design suggestions, their potential for evaluating visualizations remains under-explored. Collecting data from humans for evaluations has been a bottleneck for visualization research in terms of both time and money, and if LLMs were able to serve, even in some limited role, as evaluators, they could be a significant resource. To investigate the feasibility of using LLMs in the visualization evaluation process, we explore the extent to which LLMs possess visualization literacy-a crucial factor for their effective utility in the field. We conducted a series of experiments using a modified 53-item Visualization Literacy Assessment Test (VLAT) for and . Our findings indicate that the LLMs we explored currently fail to achieve the same levels of visualization literacy when compared to data from the general public reported in VLAT, and LLMs heavily relied on their pre-existing knowledge to answer questions instead of utilizing the information provided by the visualization when answering questions.

Modern machine learning often relies on optimizing a neural network's parameters using a loss function to learn complex features. Beyond training, examining the loss function with respect to a network's parameters (i.e., as a loss landscape) can reveal insights into the architecture and learning process. While the local structure of the loss landscape surrounding an individual solution can be characterized using a variety of approaches, the global structure of a loss landscape, which includes potentially many local minima corresponding to different solutions, remains far more difficult to conceptualize and visualize. To address this difficulty, we introduce LossLens, a visual analytics framework that explores loss landscapes at multiple scales. LossLens integrates metrics from global and local scales into a comprehensive visual representation, enhancing model diagnostics. We demonstrate LossLens through two case studies: visualizing how residual connections influence a ResNet-20, and visualizing how physical parameters influence a physics-informed neural network (PINN) solving a simple convection problem.

Nowadays, temporal heterogeneous graphs attract much research and industrial attention for building the next-generation Relational Deep Learning models and applications, due to their informative structures and features. While providing timely and precise services like personalized recommendations and question answering, this rich information also introduces extra exposure risk for each node in the graph. The distinctive local topology, the abundant heterogeneous features, and the time dimension of the graph data are more prone to expose sensitive information and narrow down the scope of victim candidates, which calls for well-defined protection techniques on graphs. To this end, we propose a Temporal Heterogeneous Graph Generator balancing Privacy, Utility, and Efficiency, named THePUff. More specifically, we first propose a differential privacy algorithm to perturb the input temporal heterogeneous graph for protecting privacy, and then utilize both the perturbed graph and the original one in a generative adversarial setting for THePUff to learn and generate privacy-guaranteed and utility-preserved graph data in an efficient manner. We further propose 6 new metrics in the temporal setting to measure heterogeneous graph utility and privacy. Finally, based on temporal heterogeneous graph datasets with up to 1 million nodes and 20 million edges, the experiments show that THePUff generates utilizable temporal heterogeneous graphs with privacy protected, compared with state-of-the-art baselines.

Accurate predictions rely on the expressiveness power of graph deep learning frameworks like graph neural networks and graph transformers, where a positional encoding mechanism has become much more indispensable in recent state-of-the-art (SOTA) works to record the canonical position information. However, the current positional encoding limits in three aspects, at least: (1) most positional encodings are pre-defined, and fixed functions, which are inadequate to adapt to the complex attributed graphs; (2) a few pioneering works propose the learnable positional encoding but still limited to the structural information, leaving the real-world time-evolving topological and feature information untouched; (3) most positional encodings should be equipped with transformer's attention mechanism to fully release the power, where the dense or relational attention is often unaffordable on large-scale structured data.Hence, we study the possibility of Learnable Spatial-Temporal Positional Encoding in an effective and efficient manner and then propose a simple temporal link prediction model named L-STEP. Briefly, for L-STEP, we (1) prove the proposed positional learning scheme can preserve the graph property from the spatial-temporal spectral viewpoint, (2) verify that MLPs can fully exploit the expressiveness and reach Transformers' performance on that encoding, (3) change different initial positional encoding inputs to show robustness, (4) analyze the theoretical complexity and obtain less empirical running time than SOTA, and (5) demonstrate its temporal link prediction out-performance on 13 classic datasets and with 10 algorithms in both transductive and inductive settings using 3 different sampling strategies. Also, L-STEP obtains the leading performance in the newest large-scale TGB benchmark.

Recent years have witnessed growing interest in understanding the sensitivity of machine learning to training data characteristics. While researchers have claimed the benefits of activities such as a human-in-the-loop approach of interactive label correction for improving model performance, there have been limited studies to quantitatively probe the relationship between the cost of label correction and the associated benefit in model performance. We employ a simulation-based approach to explore the efficacy of label correction under diverse task conditions, namely different datasets, noise properties, and machine learning algorithms. We measure the impact of label correction on model performance under the best-case scenario assumption: perfect correction (perfect human and visual systems), serving as an upper-bound estimation of the benefits derived from visual interactive label correction. The simulation results reveal a trade-off between the label correction effort expended and model performance improvement. Notably, task conditions play a crucial role in shaping the trade-off. Based on the simulation results, we develop a set of recommendations to help practitioners determine conditions under which interactive label correction is an effective mechanism for improving model performance.

Explainable Artificial Intelligence (XAI) can enhance trust in AI by offering cues that support human reasoning of AI behavior. Yet its effects on trust evolution remain unclear, especially when AI makes errors. This study examines how explanations of AI predictions influence human trust in AI-assisted decision-making under varying error severities. We tested two XAI visualizations, two AI error types, and three explanation strategies in simulated baggage screening tasks through an online study. Responses from 280 participants show that XAI representation significantly affects human compliance with AI during errors, while AI error type further shapes compliance after AI errors. AI Error type also impacts verification behaviors during AI errors, such as requesting explanations or ground truth. Moreover, strategies for conveying XAI influence perceived trust in AI, highlighting important implications for generalizing XAI effects beyond lab-based trust research.

Recent advancements in multimodal large language models (MLLMs) have driven researchers to explore how well these models read data visualizations, e.g., bar charts, scatter plots. More recently, attention has shifted to visual question answering with maps (Map-VQA). However, Map-VQA research has primarily focused on choropleth maps, which cover only a limited range of thematic categories and visual analytical tasks. To address these gaps, we introduce MapIQ, a benchmark dataset comprising 14,706 question-answer pairs across three map types: choropleth maps, cartograms, and proportional symbol maps spanning topics from six distinct themes (e.g., housing, crime). We evaluate multiple MLLMs using six visual analytical tasks, comparing their performance against one another and a human baseline. An additional experiment examining the impact of map design changes (e.g., altered color schemes, modified legend designs, and removal of map elements) provides insights into the robustness and sensitivity of MLLMs, their reliance on internal geographic knowledge, and potential avenues for improving Map-VQA performance.

This article presents a well-scaling parallel algorithm for the computation of Morse-Smale (MS) segmentations, including the region separators and region boundaries. The segmentation of the domain into ascending and descending manifolds, solely defined on the vertices, improves the computational time using path compression and fully segments the border region. Region boundaries and region separators are generated using a multi-label marching tetrahedra algorithm. This enables a fast and simple solution to find optimal parameter settings in preliminary exploration steps by generating an MS complex preview. It also poses a rapid option to generate a fast visual representation of the region geometries for immediate utilization. Two experiments demonstrate the performance of our approach with speedups of over an order of magnitude in comparison to two publicly available implementations. The example section shows the similarity to the MS complex, the useability of the approach, and the benefits of this method with respect to the presented datasets. We provide our implementation with the paper.

Machine learning models have gained traction as decision support tools for tasks that require processing copious amounts of data. However, to achieve the primary benefits of automating this part of decision-making, people must be able to trust the machine learning model's outputs. In order to enhance people's trust and promote appropriate reliance on the model, visualization techniques such as interactive model steering, performance analysis, model comparison, and uncertainty visualization have been proposed. In this study, we tested the effects of two uncertainty visualization techniques in a college admissions forecasting task, under two task difficulty levels, using Amazon's Mechanical Turk platform. Results show that (1) people's reliance on the model depends on the task difficulty and level of machine uncertainty and (2) ordinal forms of expressing model uncertainty are more likely to calibrate model usage behavior. These outcomes emphasize that reliance on decision support tools can depend on the cognitive accessibility of the visualization technique and perceptions of model performance and task difficulty.

Ensuring equitable impact of machine learning models across different societal groups is of utmost importance for real-world machine learning applications. Prior research in fairness has predominantly focused on adjusting model outputs through pre-processing, in-processing, or post-processing techniques. These techniques focus on correcting bias in either the data or the model. However, we argue that the bias in the data and model should be addressed in conjunction. To achieve this, we propose an algorithm called GroupDebias to reduce unfairness in the data in a model-guided fashion, thereby enabling models to exhibit more equitable behavior. Even though it is model-aware, the core idea of GroupDebias is independent of the model architecture, making it a versatile and effective approach that can be broadly applied across various domains and model types. Our method focuses on systematically addressing biases present in the training data itself by adaptively dropping samples that increase the biases in the model. Theoretically, the proposed approach enjoys a guaranteed improvement in demographic parity at the expense of a bounded reduction in balanced accuracy. A comprehensive evaluation of the GroupDebias algorithm through extensive experiments on diverse datasets and machine learning models demonstrates that GroupDebias consistently and significantly outperforms existing fairness enhancement techniques, achieving a more substantial reduction in unfairness with minimal impact on model performance.

We study deceptive fairness attacks on graphs to answer the following question: How can we achieve poisoning attacks on a graph learning model to exacerbate the bias deceptively? We answer this question via a bi-level optimization problem and propose a meta learning-based framework named FATE. FATE is broadly applicable with respect to various fairness definitions and graph learning models, as well as arbitrary choices of manipulation operations. We further instantiate FATE to attack statistical parity and individual fairness on graph neural networks. We conduct extensive experimental evaluations on real-world datasets in the task of semi-supervised node classification. The experimental results demonstrate that FATE could amplify the bias of graph neural networks with or without fairness consideration while maintaining the utility on the downstream task. We hope this paper provides insights into the adversarial robustness of fair graph learning and can shed light on designing robust and fair graph learning in future studies.