The State of the Art in Enhancing Trust in Machine Learning Models with the Use of Visualizations

Machine learning (ML) models are nowadays used in complex applications in various domains, such as medicine, bioinformatics, and other sciences. Due to their black box nature, however, it may sometimes be hard to understand and trust the results they provide. This has increased the demand for reliable visualization tools related to enhancing trust in ML models, which has become a prominent topic of research in the visualization community over the past decades. To provide an overview and present the frontiers of current research on the topic, we present a State-of-the-Art Report (STAR) on enhancing trust in ML models with the use of interactive visualization. We define and describe the background of the topic, introduce a categorization for visualization techniques that aim to accomplish this goal, and discuss insights and opportunities for future research directions. Among our contributions is a categorization of trust against different facets of interactive ML, expanded and improved from previous research. Our results are investigated from different analytical perspectives: (a) providing a statistical overview, (b) summarizing key findings, (c) performing topic analyses, and (d) exploring the data sets used in the individual papers, all with the support of an interactive web-based survey browser. We intend this survey to be beneficial for visualization researchers whose interests involve making ML models more trustworthy, as well as researchers and practitioners from other disciplines in their search for effective visualization techniques suitable for solving their tasks with confidence and conveying meaning to their data.     

FlexEvent: going beyond Case-Centric Exploration and Analysis of Multivariate Event Sequences

In many domains, multivariate event sequence data is collected focused around an entity (the case). Typically, each event has multiple attributes, for example, in healthcare a patient has events such as hospitalization, medication, and surgery. In addition to the multivariate events, also the case (a specific attribute, e.g., patient) has associated multivariate data (e.g., age, gender, weight). Current work typically only visualizes one attribute per event (label) in the event sequences. As a consequence, events can only be explored from a predefined case-centric perspective. However, to find complex relations from multiple perspectives (e.g., from different case definitions, such as doctor), users also need an event- and attribute-centric perspective. In addition, support is needed to effortlessly switch between and within perspectives. To support such a rich exploration, we present FlexEvent: an exploration and analysis method that enables investigation beyond a fixed case-centric perspective. Based on an adaptation of existing visualization techniques, such as scatterplots and juxtaposed small multiples, we enable flexible switching between different perspectives to explore the multivariate event sequence data needed to answer multi-perspective hypotheses. We evaluated FlexEvent with three domain experts in two use cases with sleep disorder and neonatal ICU data that show our method facilitates experts in exploring and analyzing real-world multivariate sequence data from different perspectives.     

A Visual Analytics Approach to Facilitate Crime Hotspot Analysis

Computer-based technology has played a significant role in crime prevention over the past 30 years, especially with the popularization of spatial databases and crime mapping systems. Police departments frequently use hotspot analysis to identify regions that should be a priority in receiving preventive resources. Practitioners and researchers agree that tracking crime over time and identifying its geographic patterns are vital information for planning efficiently. Frequently, police departments have access to systems that are too complicated and excessively technical, leading to modest usage. By working closely together with domain experts from police agencies of two different countries, we identified and characterized five domain tasks inherent to the hotspot analysis problem and developed SHOC, a visualization tool that strives for simplicity and ease of use in helping users to perform all the domain tasks. SHOC is included in a visual analytics system that allows users without technical expertise to annotate, save, and share analyses. We also demonstrate that our system effectively supports the completion of the domain tasks in two different real-world case studies.     

DRLViz: Understanding Decisions and Memory in Deep Reinforcement Learning

We present DRLViz, a visual analytics interface to interpret the internal memory of an agent (e.g. a robot) trained using deep reinforcement learning. This memory is composed of large temporal vectors updated when the agent moves in an environment and is not trivial to understand due to the number of dimensions, dependencies to past vectors, spatial/temporal correlations, and co-correlation between dimensions. It is often referred to as a black box as only inputs (images) and outputs (actions) are intelligible for humans. Using DRLViz, experts are assisted to interpret decisions using memory reduction interactions, and to investigate the role of parts of the memory when errors have been made (e.g. wrong direction). We report on DRLViz applied in the context of video games simulators (ViZDoom) for a navigation scenario with item gathering tasks. We also report on experts evaluation using DRLViz, and applicability of DRLViz to other scenarios and navigation problems beyond simulation games, as well as its contribution to black box models interpretability and explain-ability in the field of visual analytics.     

PAVED: Pareto Front Visualization for Engineering Design

Design problems in engineering typically involve a large solution space and several potentially conflicting criteria. Selecting a compromise solution is often supported by optimization algorithms that compute hundreds of Pareto-optimal solutions, thus informing a decision by the engineer. However, the complexity of evaluating and comparing alternatives increases with the number of criteria that need to be considered at the same time. We present a design study on Pareto front visualization to support engineers in applying their expertise and subjective preferences for selection of the most-preferred solution. We provide a characterization of data and tasks from the parametric design of electric motors. The requirements identified were the basis for our development of PAVED, an interactive parallel coordinates visualization for exploration of multi-criteria alternatives. We reflect on our user-centered design process that included iterative refinement with real data in close collaboration with a domain expert as well as a summative evaluation in the field. The results suggest a high usability of our visualization as part of a real-world engineering design workflow. Our lessons learned can serve as guidance to future visualization developers targeting multi-criteria optimization problems in engineering design or alternative domains.     

Tac-Anticipator: Visual Analytics of Anticipation Behaviors in Table Tennis Matches

Anticipation skill is important for elite racquet sports players. Successful anticipation allows them to predict the actions of the opponent better and take early actions in matches. Existing studies of anticipation behaviors, largely based on the analysis of in-lab behaviors, failed to capture the characteristics of in-situ anticipation behaviors in real matches. This research proposes a data-driven approach for research on anticipation behaviors to gain more accurate and reliable insight into anticipation skills. Collaborating with domain experts in table tennis, we develop a complete solution that includes data collection, the development of a model to evaluate anticipation behaviors, and the design of a visual analytics system called Tac-Anticipator. Our case study reveals the strengths and weaknesses of top table tennis players' anticipation behaviors. In a word, our work enriches the research methods and guidelines for visual analytics of anticipation behaviors.     

SeqDynamics: Visual Analytics for Evaluating Online Problem-solving Dynamics

Problem-solving dynamics refers to the process of solving a series of problems over time, from which a student's cognitive skills and non-cognitive traits and behaviors can be inferred. For example, we can derive a student's learning curve (an indicator of cognitive skill) from the changes in the difficulty level of problems solved, or derive a student's self-regulation patterns (an example of non-cognitive traits and behaviors) based on the problem-solving frequency over time. Few studies provide an integrated overview of both aspects by unfolding the problem-solving process. In this paper, we present a visual analytics system named SeqDynamics that evaluates students ‘problem-solving dynamics from both cognitive and non-cognitive perspectives. The system visualizes the chronological sequence of learners’ problem-solving behavior through a set of novel visual designs and coordinated contextual views, enabling users to compare and evaluate problem-solving dynamics on multiple scales. We present three scenarios to demonstrate the usefulness of SeqDynamics on a real-world dataset which consists of thousands of problem-solving traces. We also conduct five expert interviews to show that SeqDynamics enhances domain experts’ understanding of learning behavior sequences and assists them in completing evaluation tasks efficiently.     

Been There,Seen That: Visualization of Movement and 3D Eye Tracking Data from Real-World Environments

The distribution of visual attention can be evaluated using eye tracking, providing valuable insights into usability issues and interaction patterns. However, when used in real, augmented, and collaborative environments, new challenges arise that go beyond desktop scenarios and purely virtual environments. Toward addressing these challenges, we present a visualization technique that provides complementary views on the movement and eye tracking data recorded from multiple people in real-world environments. Our method is based on a space-time cube visualization and a linked 3D replay of recorded data. We showcase our approach with an experiment that examines how people investigate an artwork collection. The visualization provides insights into how people moved and inspected individual pictures in their spatial context over time. In contrast to existing methods, this analysis is possible for multiple participants without extensive annotation of areas of interest. Our technique was evaluated with a think-aloud experiment to investigate analysis strategies and an interview with domain experts to examine the applicability in other research fields.     

VA-TRAC: Geospatial Trajectory Analysis for Monitoring,Identification, and Verification in Fishing Vessel Operations

In order to ensure sustainability, fishing operations are governed by many rules and regulations that restrict the use of certain techniques and equipment, specify the species and size offish that can be harvested, and regulate commercial activities based on licensing schemes. As the world's second largest exporter offish and seafood products, Norway invests a significant amount of effort into maintaining natural ecosystem dynamics by ensuring compliance with its constantly evolving science-based regulatory body. This paper introduces VA-TRAC, a geovisual analytics application developed in collaboration with the Norwegian Directorate of Fisheries in order to address this complex task. Our approach uses automatic methods to identify possible catch operations based on fishing vessel trajectories, embedded in an interactive web-based visual interface used to explore the results, compare them with licensing information, and incorporate the analysts’ domain knowledge into the decision making process. We present a data and task analysis based on a close collaboration with domain experts, and the design and implementation of VA-TRAC to address the identified requirements.     

Co-creating Visualizations: A First Evaluation with Social Science Researchers

Co-creation is a design method where designers and domain experts work together to develop a product. In this paper, we present and evaluate the use of co-creation to design a visual information system with social science researchers in order to explore and analyze their data. Co-creation proposes involving the future users in the design process to ensure that they play a critical role in the design, and to increase the chances of long-term adoption. We evaluated the co-creation process through surveys, interviews and a user study. According to the participants’ feedback, they felt listened to through co-creation, and considered the methodology helpful to develop visualizations that support their research in the near future. However, participation was far from perfect, particularly early career researchers showed limited interest in participating because they did not see the process as beneficial for their research publication goals. We summarize benefits and limitations of co-creation, together with our recommendations, as lessons learned.     

A Laplacian for Nonmanifold Triangle Meshes

We describe a discrete Laplacian suitable for any triangle mesh, including those that are nonmanifold or nonorientable (with or without boundary). Our Laplacian is a robust drop-in replacement for the usual cotan matrix, and is guaranteed to have nonnegative edge weights on both interior and boundary edges, even for extremely poor-quality meshes. The key idea is to build what we call a “tufted cover” over the input domain, which has nonmanifold vertices but manifold edges. Since all edges are manifold, we can flip to an intrinsic Delaunay triangulation; our Laplacian is then the cotan Laplacian of this new triangulation. This construction also provides a high-quality point cloud Laplacian, via a nonmanifold triangulation of the point set. We validate our Laplacian on a variety of challenging examples (including all models from Thingi10k), and a variety of standard tasks including geodesic distance computation, surface deformation, parameterization, and computing minimal surfaces.     

Infomages: Embedding Data into Thematic Images

Recent studies have indicated that visually embellished charts such as infographics have the ability to engage viewers and positively affect memorability. Fueled by these findings, researchers have proposed a variety of infographic design tools. However, these tools do not cover the entire design space. In this work, we identify a subset of infographics that we call infomages. Infomages are casual visuals of data in which a data chart is embedded into a thematic image such that the content of the image reflects the subject and the designer's interpretation of the data. Creating an effective infomage, however, can require a fair amount of design expertise and is thus out of reach for most people. In order to also afford non-artists with the means to design convincing infomages, we first study the principled design of existing infomages and identify a set of key chart embedding techniques. Informed by these findings we build a design tool that links web-scale image search with a set of interactive image processing tools to empower novice users with the ability to design a wide variety of infomages. As the embedding process might introduce some amount of visual distortion of the data our tool also aids users to gauge the amount of this distortion, if any. We experimentally demonstrate the usability of our tool and conclude with a discussion of infomages and our design tool.     

VISITOR: Visual Interactive State Sequence Exploration for Reinforcement Learning

Understanding the behavior of deep reinforcement learning agents is a crucial requirement throughout their development. Existing work has addressed the identification of observable behavioral patterns in state sequences or analysis of isolated internal representations; however, the overall decision-making of deep-learning RL agents remains opaque. To tackle this, we present VISITOR, a visual analytics system enabling the analysis of entire state sequences, the diagnosis of singular predictions, and the comparison between agents. A sequence embedding view enables the multiscale analysis of state sequences, utilizing custom embedding techniques for a stable spatialization of the observations and internal states. We provide multiple layers: (1) a state space embedding, highlighting different groups of states inside the state-action sequences, (2) a trajectory view, emphasizing decision points, (3) a network activation mapping, visualizing the relationship between observations and network activations, (4) a transition embedding, enabling the analysis of state-to-state transitions. The embedding view is accompanied by an interactive reward view that captures the temporal development of metrics, which can be linked directly to states in the embedding. Lastly, a model list allows for the quick comparison of models across multiple metrics. Annotations can be exported to communicate results to different audiences. Our two-stage evaluation with eight experts confirms the effectiveness in identifying states of interest, comparing the quality of policies, and reasoning about the internal decision-making processes.     

Topology-Aware Surface Reconstruction for Point Clouds

We present an approach to incorporate topological priors in the reconstruction of a surface from a point scan. We base the reconstruction on basis functions which are optimized to provide a good fit to the point scan while satisfying predefined topological constraints. We optimize the parameters of a model to obtain a likelihood function over the reconstruction domain. The topological constraints are captured by persistence diagrams which are incorporated within the optimization algorithm to promote the correct topology. The result is a novel topology-aware technique which can (i) weed out topological noise from point scans, and (ii) capture certain nuanced properties of the underlying shape which could otherwise be lost while performing surface reconstruction. We show results reconstructing shapes with multiple potential topologies, compare to other classical surface construction techniques, and show the completion of real scan data.     

QUESTO: Interactive Construction of Objective Functions for Classification Tasks

Building effective classifiers requires providing the modeling algorithms with information about the training data and modeling goals in order to create a model that makes proper tradeoffs. Machine learning algorithms allow for flexible specification of such meta-information through the design of the objective functions that they solve. However, such objective functions are hard for users to specify as they are a specific mathematical formulation of their intents. In this paper, we present an approach that allows users to generate objective functions for classification problems through an interactive visual interface. Our approach adopts a semantic interaction design in that user interactions over data elements in the visualization are translated into objective function terms. The generated objective functions are solved by a machine learning solver that provides candidate models, which can be inspected by the user, and used to suggest refinements to the specifications. We demonstrate a visual analytics system QUESTO for users to manipulate objective functions to define domain-specific constraints. Through a user study we show that QUESTO helps users create various objective functions that satisfy their goals.     

Memento: Localized Time-Warping for Spatio-Temporal Selection

Interaction techniques for temporal data are often focused on affecting the spatial aspects of the data, for instance through the use of transfer functions, camera navigation or clipping planes. However, the temporal aspect of the data interaction is often neglected. The temporal component is either visualized as individual time steps, an animation or a static summary over the temporal domain. When dealing with streaming data, these techniques are unable to cope with the task of re-viewing an interesting local spatio-temporal event, while continuing to observe the rest of the feed. We propose a novel technique that allows users to interactively specify areas of interest in the spatio-temporal domain. By employing a time-warp function, we are able to slow down time, freeze time or even travel back in time, around spatio-temporal events of interest. The combination of such a (pre-defined) time-warp function and brushing directly in the data to select regions of interest allows for a detailed review of temporally and spatially localized events, while maintaining an overview of the global spatio-temporal data. We demonstrate the utility of our technique with several usage scenarios.     

GTMapLens: Interactive Lens for Geo-Text Data Browsing on Map

Data containing geospatial semantics, such as geotagged tweets, travel blogs, and crime reports, associates natural language texts with geographical locations. This paper presents a lens-based visual interaction technique, GTMapLens, to flexibly browse the geo-text data on a map. It allows users to perform dynamic focus+context exploration by using movable lenses to browse geographical regions, find locations of interest, and perform comparative and drill-down studies. Geo-text data is visualized in a way that users can easily perceive the underlying geospatial semantics along with lens moving. Based on a requirement analysis with a cohort of multidisciplinary domain experts, a set of lens interaction techniques are developed including keywords control, path management, context visualization, and snapshot anchors. They allow users to achieve a guided and controllable exploration of geo-text data. A hierarchical data model enables the interactive lens operations by accelerated data retrieval from a geo-text database. Evaluation with real-world datasets is presented to show the usability and effectiveness of GTMapLens.