ProtoSteer: Steering Deep Sequence Model with Prototypes

CompositingVis: Exploring Interactions for Creating Composite Visualizations in Immersive Environments

Composite visualization represents a widely embraced design that combines multiple visual representations to create an integrated view. However, the traditional approach of creating composite visualizations in immersive environments typically occurs asynchronously outside of the immersive space and is carried out by experienced experts. In this work, we aim to empower users to participate in the creation of composite visualization within immersive environments through embodied interactions. This could provide a flexible and fluid experience with immersive visualization and has the potential to facilitate understanding of the relationship between visualization views. We begin with developing a design space of embodied interactions to create various types of composite visualizations with the consideration of data relationships. Drawing inspiration from people's natural experience of manipulating physical objects, we design interactions based on the combination of 3D manipulations in immersive environments. Building upon the design space, we present a series of case studies showcasing the interaction to create different kinds of composite visualizations in virtual reality. Subsequently, we conduct a user study to evaluate the usability of the derived interaction techniques and user experience of creating composite visualizations through embodied interactions. We find that empowering users to participate in composite visualizations through embodied interactions enables them to flexibly leverage different visualization views for understanding and communicating the relationships between different views, which underscores the potential of several future application scenarios.

AdversaFlow: Visual Red Teaming for Large Language Models with Multi-Level Adversarial Flow

Large Language Models (LLMs) are powerful but also raise significant security concerns, particularly regarding the harm they can cause, such as generating fake news that manipulates public opinion on social media and providing responses to unethical activities. Traditional red teaming approaches for identifying AI vulnerabilities rely on manual prompt construction and expertise. This paper introduces AdversaFlow, a novel visual analytics system designed to enhance LLM security against adversarial attacks through human-AI collaboration. AdversaFlow involves adversarial training between a target model and a red model, featuring unique multi-level adversarial flow and fluctuation path visualizations. These features provide insights into adversarial dynamics and LLM robustness, enabling experts to identify and mitigate vulnerabilities effectively. We present quantitative evaluations and case studies validating our system's utility and offering insights for future AI security solutions. Our method can enhance LLM security, supporting downstream scenarios like social media regulation by enabling more effective detection, monitoring, and mitigation of harmful content and behaviors.

A Multi-Level Task Framework for Event Sequence Analysis

Despite the development of numerous visual analytics tools for event sequence data across various domains, including but not limited to healthcare, digital marketing, and user behavior analysis, comparing these domain-specific investigations and transferring the results to new datasets and problem areas remain challenging. Task abstractions can help us go beyond domain-specific details, but existing visualization task abstractions are insufficient for event sequence visual analytics because they primarily focus on multivariate datasets and often overlook automated analytical techniques. To address this gap, we propose a domain-agnostic multi-level task framework for event sequence analytics, derived from an analysis of 58 papers that present event sequence visualization systems. Our framework consists of four levels: objective, intent, strategy, and technique. Overall objectives identify the main goals of analysis. Intents comprises five high-level approaches adopted at each analysis step: augment data, simplify data, configure data, configure visualization, and manage provenance. Each intent is accomplished through a number of strategies, for instance, data simplification can be achieved through aggregation, summarization, or segmentation. Finally, each strategy can be implemented by a set of techniques depending on the input and output components. We further show that each technique can be expressed through a quartet of action-input-output-criteria. We demonstrate the framework's descriptive power through case studies and discuss its similarities and differences with previous event sequence task taxonomies.

Visual Analytics for Efficient Image Exploration and User-Guided Image Captioning

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

DASS Good: Explainable Data Mining of Spatial Cohort Data

Developing applicable clinical machine learning models is a difficult task when the data includes spatial information, for example, radiation dose distributions across adjacent organs at risk. We describe the co-design of a modeling system, DASS, to support the hybrid human-machine development and validation of predictive models for estimating long-term toxicities related to radiotherapy doses in head and neck cancer patients. Developed in collaboration with domain experts in oncology and data mining, DASS incorporates human-in-the-loop visual steering, spatial data, and explainable AI to augment domain knowledge with automatic data mining. We demonstrate DASS with the development of two practical clinical stratification models and report feedback from domain experts. Finally, we describe the design lessons learned from this collaborative experience.

Polyphony: an Interactive Transfer Learning Framework for Single-Cell Data Analysis

Reference-based cell-type annotation can significantly reduce time and effort in single-cell analysis by transferring labels from a previously-annotated dataset to a new dataset. However, label transfer by end-to-end computational methods is challenging due to the entanglement of technical (e.g., from different sequencing batches or techniques) and biological (e.g., from different cellular microenvironments) variations, only the first of which must be removed. To address this issue, we propose Polyphony, an interactive transfer learning (ITL) framework, to complement biologists' knowledge with advanced computational methods. Polyphony is motivated and guided by domain experts' needs for a controllable, interactive, and algorithm-assisted annotation process, identified through interviews with seven biologists. We introduce anchors, i.e., analogous cell populations across datasets, as a paradigm to explain the computational process and collect user feedback for model improvement. We further design a set of visualizations and interactions to empower users to add, delete, or modify anchors, resulting in refined cell type annotations. The effectiveness of this approach is demonstrated through quantitative experiments, two hypothetical use cases, and interviews with two biologists. The results show that our anchor-based ITL method takes advantage of both human and machine intelligence in annotating massive single-cell datasets.

RASIPAM: Interactive Pattern Mining of Multivariate Event Sequences in Racket Sports

Experts in racket sports like tennis and badminton use tactical analysis to gain insight into competitors' playing styles. Many data-driven methods apply pattern mining to racket sports data - which is often recorded as multivariate event sequences - to uncover sports tactics. However, tactics obtained in this way are often inconsistent with those deduced by experts through their domain knowledge, which can be confusing to those experts. This work introduces RASIPAM, a RAcket-Sports Interactive PAttern Mining system, which allows experts to incorporate their knowledge into data mining algorithms to discover meaningful tactics interactively. RASIPAM consists of a constraint-based pattern mining algorithm that responds to the analysis demands of experts: Experts provide suggestions for finding tactics in intuitive written language, and these suggestions are translated into constraints to run the algorithm. RASIPAM further introduces a tailored visual interface that allows experts to compare the new tactics with the original ones and decide whether to apply a given adjustment. This interactive workflow iteratively progresses until experts are satisfied with all tactics. We conduct a quantitative experiment to show that our algorithm supports real-time interaction. Two case studies in tennis and in badminton respectively, each involving two domain experts, are conducted to show the effectiveness and usefulness of the system.

FeatureEnVi: Visual Analytics for Feature Engineering Using Stepwise Selection and Semi-Automatic Extraction Approaches

The machine learning (ML) life cycle involves a series of iterative steps, from the effective gathering and preparation of the data-including complex feature engineering processes-to the presentation and improvement of results, with various algorithms to choose from in every step. Feature engineering in particular can be very beneficial for ML, leading to numerous improvements such as boosting the predictive results, decreasing computational times, reducing excessive noise, and increasing the transparency behind the decisions taken during the training. Despite that, while several visual analytics tools exist to monitor and control the different stages of the ML life cycle (especially those related to data and algorithms), feature engineering support remains inadequate. In this paper, we present FeatureEnVi, a visual analytics system specifically designed to assist with the feature engineering process. Our proposed system helps users to choose the most important feature, to transform the original features into powerful alternatives, and to experiment with different feature generation combinations. Additionally, data space slicing allows users to explore the impact of features on both local and global scales. FeatureEnVi utilizes multiple automatic feature selection techniques; furthermore, it visually guides users with statistical evidence about the influence of each feature (or subsets of features). The final outcome is the extraction of heavily engineered features, evaluated by multiple validation metrics. The usefulness and applicability of FeatureEnVi are demonstrated with two use cases and a case study. We also report feedback from interviews with two ML experts and a visualization researcher who assessed the effectiveness of our system.

GenNI: Human-AI Collaboration for Data-Backed Text Generation

Table2Text systems generate textual output based on structured data utilizing machine learning. These systems are essential for fluent natural language interfaces in tools such as virtual assistants; however, left to generate freely these ML systems often produce misleading or unexpected outputs. GenNI (Generation Negotiation Interface) is an interactive visual system for high-level human-AI collaboration in producing descriptive text. The tool utilizes a deep learning model designed with explicit control states. These controls allow users to globally constrain model generations, without sacrificing the representation power of the deep learning models. The visual interface makes it possible for users to interact with AI systems following a Refine-Forecast paradigm to ensure that the generation system acts in a manner human users find suitable. We report multiple use cases on two experiments that improve over uncontrolled generation approaches, while at the same time providing fine-grained control. A demo and source code are available at https://genni.vizhub.ai.

Towards Visual Explainable Active Learning for Zero-Shot Classification

Zero-shot classification is a promising paradigm to solve an applicable problem when the training classes and test classes are disjoint. Achieving this usually needs experts to externalize their domain knowledge by manually specifying a class-attribute matrix to define which classes have which attributes. Designing a suitable class-attribute matrix is the key to the subsequent procedure, but this design process is tedious and trial-and-error with no guidance. This paper proposes a visual explainable active learning approach with its design and implementation called semantic navigator to solve the above problems. This approach promotes human-AI teaming with four actions (ask, explain, recommend, respond) in each interaction loop. The machine asks contrastive questions to guide humans in the thinking process of attributes. A novel visualization called semantic map explains the current status of the machine. Therefore analysts can better understand why the machine misclassifies objects. Moreover, the machine recommends the labels of classes for each attribute to ease the labeling burden. Finally, humans can steer the model by modifying the labels interactively, and the machine adjusts its recommendations. The visual explainable active learning approach improves humans' efficiency of building zero-shot classification models interactively, compared with the method without guidance. We justify our results with user studies using the standard benchmarks for zero-shot classification.

VizSnippets: Compressing Visualization Bundles Into Representative Previews for Browsing Visualization Collections

Visualization collections, accessed by platforms such as Tableau Online or Power Bl, are used by millions of people to share and access diverse analytical knowledge in the form of interactive visualization bundles. Result snippets, compact previews of these bundles, are presented to users to help them identify relevant content when browsing collections. Our engagement with Tableau product teams and review of existing snippet designs on five platforms showed us that current practices fail to help people judge the relevance of bundles because they include only the title and one image. Users frequently need to undertake the time-consuming endeavour of opening a bundle within its visualization system to examine its many views and dashboards. In response, we contribute the first systematic approach to visualization snippet design. We propose a framework for snippet design that addresses eight key challenges that we identify. We present a computational pipeline to compress the visual and textual content of bundles into representative previews that is adaptive to a provided pixel budget and provides high information density with multiple images and carefully chosen keywords. We also reflect on the method of visual inspection through random sampling to gain confidence in model and parameter choices.

Seek for Success: A Visualization Approach for Understanding the Dynamics of Academic Careers

How to achieve academic career success has been a long-standing research question in social science research. With the growing availability of large-scale well-documented academic profiles and career trajectories, scholarly interest in career success has been reinvigorated, which has emerged to be an active research domain called the Science of Science (i.e., SciSci). In this study, we adopt an innovative dynamic perspective to examine how individual and social factors will influence career success over time. We propose ACSeeker, an interactive visual analytics approach to explore the potential factors of success and how the influence of multiple factors changes at different stages of academic careers. We first applied a Multi-factor Impact Analysis framework to estimate the effect of different factors on academic career success over time. We then developed a visual analytics system to understand the dynamic effects interactively. A novel timeline is designed to reveal and compare the factor impacts based on the whole population. A customized career line showing the individual career development is provided to allow a detailed inspection. To validate the effectiveness and usability of ACSeeker, we report two case studies and interviews with a social scientist and general researchers.

Overview of Explainable Artificial Intelligence for Prognostic and Health Management of Industrial Assets Based on Preferred Reporting Items for Systematic Reviews and Meta-Analyses

Surveys on explainable artificial intelligence (XAI) are related to biology, clinical trials, fintech management, medicine, neurorobotics, and psychology, among others. Prognostics and health management (PHM) is the discipline that links the studies of failure mechanisms to system lifecycle management. There is a need, which is still absent, to produce an analytical compilation of PHM-XAI works. In this paper, we use preferred reporting items for systematic reviews and meta-analyses (PRISMA) to present a state of the art on XAI applied to PHM of industrial assets. This work provides an overview of the trend of XAI in PHM and answers the question of accuracy versus explainability, considering the extent of human involvement, explanation assessment, and uncertainty quantification in this topic. Research articles associated with the subject, since 2015 to 2021, were selected from five databases following the PRISMA methodology, several of them related to sensors. The data were extracted from selected articles and examined obtaining diverse findings that were synthesized as follows. First, while the discipline is still young, the analysis indicates a growing acceptance of XAI in PHM. Second, XAI offers dual advantages, where it is assimilated as a tool to execute PHM tasks and explain diagnostic and anomaly detection activities, implying a real need for XAI in PHM. Third, the review shows that PHM-XAI papers provide interesting results, suggesting that the PHM performance is unaffected by the XAI. Fourth, human role, evaluation metrics, and uncertainty management are areas requiring further attention by the PHM community. Adequate assessment metrics to cater to PHM needs are requested. Finally, most case studies featured in the considered articles are based on real industrial data, and some of them are related to sensors, showing that the available PHM-XAI blends solve real-world challenges, increasing the confidence in the artificial intelligence models' adoption in the industry.

StackGenVis: Alignment of Data, Algorithms, and Models for Stacking Ensemble Learning Using Performance Metrics

In machine learning (ML), ensemble methods-such as bagging, boosting, and stacking-are widely-established approaches that regularly achieve top-notch predictive performance. Stacking (also called "stacked generalization") is an ensemble method that combines heterogeneous base models, arranged in at least one layer, and then employs another metamodel to summarize the predictions of those models. Although it may be a highly-effective approach for increasing the predictive performance of ML, generating a stack of models from scratch can be a cumbersome trial-and-error process. This challenge stems from the enormous space of available solutions, with different sets of data instances and features that could be used for training, several algorithms to choose from, and instantiations of these algorithms using diverse parameters (i.e., models) that perform differently according to various metrics. In this work, we present a knowledge generation model, which supports ensemble learning with the use of visualization, and a visual analytics system for stacked generalization. Our system, StackGenVis, assists users in dynamically adapting performance metrics, managing data instances, selecting the most important features for a given data set, choosing a set of top-performant and diverse algorithms, and measuring the predictive performance. In consequence, our proposed tool helps users to decide between distinct models and to reduce the complexity of the resulting stack by removing overpromising and underperforming models. The applicability and effectiveness of StackGenVis are demonstrated with two use cases: a real-world healthcare data set and a collection of data related to sentiment/stance detection in texts. Finally, the tool has been evaluated through interviews with three ML experts.