Computer Graphics Forum最新文献

Explaining Machine Learning (ML) — and especially Deep Learning (DL) — classifiers' decisions is a subject of interest across fields due to the increasing ubiquity of such models in computing systems. As models get increasingly complex, relying on sophisticated machinery to recognize data patterns, explaining their behavior becomes more difficult. Directly visualizing classifier behavior is in general infeasible, as they create partitions of the data space, which is typically high dimensional. In recent years, Decision Boundary Maps (DBMs) have been developed, taking advantage of projection and inverse projection techniques. By being able to map 2D points back to the data space and subsequently run a classifier, DBMs represent a slice of classifier outputs. However, we recognize that DBMs without additional explanatory views are limited in their applicability. In this work, we propose augmenting the naive DBM generating process with views that provide more in-depth information about classifier behavior, such as whether the training procedure is locally stable. We describe our proposed views — which we term Differentiable Decision Boundary Maps — over a running example, explaining how our work enables drawing new and useful conclusions from these dense maps. We further demonstrate the value of these conclusions by showing how useful they would be in carrying out or preventing a dataset poisoning attack. We thus provide evidence of the ability of our proposed views to make DBMs significantly more trustworthy and interpretable, increasing their utility as a model understanding tool.

In natural language processing (NLP), text classification tasks are increasingly fine-grained, as datasets are fragmented into a larger number of classes that are more difficult to differentiate from one another. As a consequence, the semantic structures of datasets have become more complex, and model decisions more difficult to explain. Existing tools, suited for coarse-grained classification, falter under these additional challenges. In response to this gap, we worked closely with NLP domain experts in an iterative design-and-evaluation process to characterize and tackle the growing requirements in their workflow of developing fine-grained text classification models. The result of this collaboration is the development of SemLa, a novel Visual Analytics system tailored for 1) dissecting complex semantic structures in a dataset when it is spatialized in model embedding space, and 2) visualizing fine-grained nuances in the meaning of text samples to faithfully explain model reasoning. This paper details the iterative design study and the resulting innovations featured in SemLa. The final design allows contrastive analysis at different levels by unearthing lexical and conceptual patterns including biases and artifacts in data. Expert feedback on our final design and case studies confirm that SemLa is a useful tool for supporting model validation and debugging as well as data annotation.

Temporal trees are trees that evolve over a discrete set of time steps. Each time step is associated with a node-weighted rooted tree and consecutive trees change by adding new nodes, removing nodes, splitting nodes, merging nodes, and changing node weights. Recently, two-dimensional visualizations of temporal trees called temporal treemaps have been proposed, representing the temporal dimension on the x-axis, and visualizing the tree modifications over time as temporal edges of varying thickness. The tree hierarchy at each time step is depicted as a vertical, one-dimensional nesting relationships, similarly to standard, non-temporal treemaps. Naturally, temporal edges can cross in the visualization, decreasing readability. Heuristics were proposed to minimize such crossings in the literature, but a formal characterization and minimization of crossings in temporal treemaps was left open. In this paper, we propose two variants of defining crossings in temporal treemaps that can be combinatorially characterized. For each variant, we propose an exact optimization algorithm based on integer linear programming and heuristics based on graph drawing techniques. In an extensive experimental evaluation, we show that on the one hand the exact algorithms reduce the number of crossings by a factor of 20 on average compared to the previous algorithms. On the other hand, our new heuristics are faster by a factor of more than 100 and still reduce the number of crossings by a factor of almost three.

The evolution and accumulation of snow cover are among the most important characteristics influencing Antarctica's climate and biotopes. The changes in Antarctica are also substantially impacting global climate change. Therefore, detailed monitoring of snow evolution is key to understanding such changes. One way to conduct this monitoring is by installing trail cameras in a particular region and then processing the captured information. This option is affordable, but has some drawbacks, such as the fully automatic solution for the extraction of snow height from these images is not feasible. Therefore, it still requires human intervention, manually correcting the inaccurately extracted information. In this paper, we present Antarstick, a tool for visual guidance of the user to potentially wrong values extracted from poor-quality images and support for their interactive correction. This tool allows for much quicker and semi-automated processing of snow height from time-lapse photography.

The research communities studying visualization and sonification for data display and analysis share exceptionally similar goals, essentially making data of any kind interpretable to humans. One community does so by using visual representations of data, and the other community employs auditory (non-speech) representations of data. While the two communities have a lot in common, they developed mostly in parallel over the course of the last few decades. With this STAR, we discuss a collection of work that bridges the borders of the two communities, hence a collection of work that aims to integrate the two techniques into one form of audiovisual display, which we argue to be “more than the sum of the two.” We introduce and motivate a classification system applicable to such audiovisual displays and categorize a corpus of 57 academic publications that appeared between 2011 and 2023 in categories such as reading level, dataset type, or evaluation system, to mention a few. The corpus also enables a meta-analysis of the field, including regularly occurring design patterns such as type of visualization and sonification techniques, or the use of visual and auditory channels, showing an overall diverse field with different designs. An analysis of a co-author network of the field shows individual teams without many interconnections. The body of work covered in this STAR also relates to three adjacent topics: audiovisual monitoring, accessibility, and audiovisual data art. These three topics are discussed individually in addition to the systematically conducted part of this research. The findings of this report may be used by researchers from both fields to understand the potentials and challenges of such integrated designs while hopefully inspiring them to collaborate with experts from the respective other field.

Charts remain widely inaccessible on the web for users of assistive technologies like screen readers. This is, in part, due to data visualization experts still lacking the experience, knowledge, and time to consistently implement accessible charts. As a result, screen reader users are prevented from accessing information and are forced to resort to tabular alternatives (if available), limiting the insights that they can gather. We worked with both groups to develop AutoVizuA11y, a tool that automates the addition of accessible features to web-based charts. It generates human-like descriptions of the data using a large language model, calculates statistical insights from the data, and provides keyboard navigation between multiple charts and underlying elements. Fifteen screen reader users interacted with charts made accessible with AutoVizuA11y in a usability test, thirteen of which praised the tool for its intuitive design, short learning curve, and rich information. On average, they took 66 seconds to complete each of the eight analytical tasks presented and achieved a success rate of 89%. Through a SUS questionnaire, the participants gave AutoVizuA11y an “Excellent” score — 83.5/100 points. We also gathered feedback from two data visualization experts who used the tool. They praised the tool availability, ease of use and functionalities, and provided feedback to add AutoVizuA11y support for other technologies in the future.

Atmospheric rivers (ARs) are long, narrow regions of water vapor in the Earth's atmosphere that transport heat and moisture from the tropics to the mid-latitudes. ARs are often associated with extreme weather events in North America and contribute significantly to water supply and flood risk. However, characterizing ARs has been a major challenge due to the lack of a universal definition and their structural variations. Existing AR detection tools (ARDTs) produce distinct AR boundaries for the same event, making the risk assessment of ARs a difficult task. Understanding these uncertainties is crucial to improving the predictability of AR impacts, including their landfall areas and associated precipitation, which could cause catastrophic flooding and landslides over the coastal regions. In this work, we develop an uncertainty visualization framework that captures boundary and interior uncertainties, i.e., structural variations, of an ensemble of ARs that arise from a set of ARDTs. We first provide a statistical overview of the AR boundaries using the contour boxplots of Whitaker et al. that highlight the structural variations of AR boundaries based on their nesting relationships. We then introduce the topological skeletons of ARs based on Morse complexes that characterize the interior variation of an ensemble of ARs. We propose an uncertainty visualization of these topological skeletons, inspired by MetroSets of Jacobson et al. that emphasizes the agreements and disagreements across the ensemble members. Through case studies and expert feedback, we demonstrate that the two approaches complement each other, and together they could facilitate an effective comparative analysis process and provide a more confident outlook on an AR's shape, area, and onshore impact.

User evaluation is a common and useful tool for systematically generating knowledge and validating novel approaches in the domain of Immersive Analytics. Since this research domain centres around users, user evaluation is of extraordinary relevance. Additionally, Immersive Analytics is an interdisciplinary field of research where different communities bring in their own methodologies. It is vital to investigate and synchronise these different approaches with the long-term goal to reach a shared evaluation framework. While there have been several studies focusing on Immersive Analytics as a whole or on certain aspects of the domain, this is the first systematic review of the state of evaluation methodology in Immersive Analytics. The main objective of this systematic literature review is to illustrate methodologies and research areas that are still underrepresented in user studies by identifying current practice in user evaluation in the domain of Immersive Analytics in coherence with the PRISMA protocol. (see https://www.acm.org/publications/class-2012)