Search

International audience; We present Moliverse, an integration of the molecular visualization framework VIAMD into the astronomical visualization software OpenSpace, allowing us to bridge the two extreme ends of the scale spectrum to show, for example, the gas composition in a planet's atmosphere or molecular structures in comet trails and can empower the creation of educational exhibitions. For that purpose we do not use a linear scale traversal but break the scale continuity and show molecular simulations as focus in the context of celestial bodies. We demonstrate the application of our concept in two storytelling scenarios and envision the application both for science presentations to lay audiences and for dedicated exploration, potentially also in a molecule-only environment.

Recent work identified a concerning trend of disproportional gender representation in research participants in Human-Computer Interaction (HCI). Motivated by the fact that Human-Robot Interaction (HRI) shares many participant practices with HCI, we explored whether this trend is mirrored in our field. By producing a dataset covering participant gender representation in all 684 full papers published at the HRI conference from 2006-2021, we identify current trends in HRI research participation. We find an over-representation of men in research participants to date, as well as inconsistent and/or incomplete gender reporting which typically engages in a binary treatment of gender at odds with published best practice guidelines. We further examine if and how participant gender has been considered in user studies to date, in-line with current discourse surrounding the importance and/or potential risks of gender based analyses. Finally, we complement this with a survey of HRI researchers to examine correlations between the who is doing with the who is taking part, to further reflect on factors which seemingly influence gender bias in research participation across different sub-fields of HRI. Through our analysis we identify areas for improvement, but also reason for optimism, and derive some practical suggestions for HRI researchers going forward. CC BY 4.0*All authors contributed equally to this work. KW has taken on first author responsibilities whilst 2nd-4th author ordering was decided by dice roll. Authors’ addresses: Katie Winkle*, Uppsala Universitet, Sweden, katie.winkle@it.uu.se; Erik Lagerstedt*, University of Skövde, Sweden,erik.lagerstedt@his.se; Ilaria Torre*, KTH Royal Institute of Technology, Sweden, ilariat@kth.se; Anna Ofenwanger*, Université Paris-Saclay, CNRS, Inria, LISN, Orsay, France, anna.ofenwanger@universite-paris-saclay.fr.We first want to acknowledge those authors who came before us in identifying issues around gender in HCI – these works have signicantly informed this manuscript, but also our approaches to HRI research more broadly. We wish to specically thank Dongwook Yoon and Julia Bullard for their work on the conceptualisation and development of the gender data schema and extraction method that we build on [59]; also Minsuk Chang, Alan Milligan, and Austin Kobayashi for their input and work on initial version of the MAGDA tool [59].We would like to thank all of the HRI researchers who engaged with our survey. We additionally want to thank those reviewers who provided constructive feedback on an earlier version of this work, and reviewersfor/attendees of the DEI Workshop held at HRI 2022 for further discussion – the manuscript is improved greatly as a result and we hope we have done justice to both the positive and negative critique we have received. All figures were generated with a colour palette based on the non-binary pride lag, created by Joel Le Forestier (https://joelleforestier.com/#pridepalettes). This work was partially funded by the Digital Futures Research Centre.

International audience; Aims: This work aimed to study the effect of noninvasive vagus nerve stimulation on severe restless legs syndrome (RLS) resistant to pharmacotherapy. Materials and Methods: Patients with severe pharmacoresistant RLS were recruited from a tertiary care sleep center. Intervention was one-hour weekly sessions of transauricular vagus nerve stimulation (tVNS) in the left cymba concha, for eight weeks. The primary outcome measure was the score on the International Restless Legs Rating Scale (IRLS); secondary outcome measures were quality of life (Restless Legs Syndrome Quality of Life scale [RLSQOL]), mood disorders using the Hospital Anxiety and Depression scale subscale for depression (HADD) and Hospital Anxiety and Depression scale subscale for anxiety (HADA), and objective sleep latency, sleep duration, efficiency, and leg movement time measured by actigraphy. Results: Fifteen patients, 53% male, aged mean 62.7 ± 12.3 years with severe RLS, reduced quality of life, and symptoms of anxiety and depression, were included. The IRLS improved from baseline to session eight: IRLS 31.9 ± 2.9 vs 24.6 ± 5.9 p = 0.0003. Of these participants, 27% (4/15) had a total response with a decrease below an IRLS score of 20; 40% (6/15) a partial response with an improvement in the IRLS > 5 but an IRLS above 20; and 33% (5/15) were nonresponders. After tVNS, quality of life improved (RLSQOL 49.3 ± 18.1 vs 80.0 ± 19.6 p = 0.0005), as did anxiety (HADA 8.9 ± 5.4 vs 6.2 ± 5.0 p = 0.001) and depression (HADD 5.2 ± 4.5 vs 4.0 ± 4.0 p = 0.01). No significant change was found in actigraphic outcome measures. Conclusions: In this pilot study, tVNS improved the symptoms of RLS in 66% of participants (10/15) with severe pharmacoresistant RLS, with concomitant improvements in quality of life and mood. Randomized controlled trials evaluating therapeutic efficacy of tVNS in RLS are needed to confirm these promising findings.

International audience; Accurate classification of cognitive states from Electroencephalographic (EEG) signals is crucial in neuroscience applications such as Brain-Computer Interfaces (BCIs). Classification pipelines based on Riemannian geometry are often state-of-the-art in the BCI field. In this type of BCI, covariance matrices based on EEG signals of independent frequency bands are used as classification features. However, there is significant neuroscience evidence of neural interactions across frequency bands, such as cross-frequency coupling (CFC). Therefore, in this paper, we propose novel symmetric positive definite (SPD) matrix representations considering CFC for Riemannian geometry-based EEG classification. The spatial interactions of phase and amplitude within and between frequency bands are described in three different CFC SPD matrices. This allows us to include additional discriminative neurophysiological features that are not available in the conventional Riemannian EEG features. Our method was evaluated using a mental workload classification task from a public passive BCI dataset. Our fused model of the three CFC covariance matrices showed statistically significant improvements in average classification accuracies from the conventional covariance matrix in the theta and alpha bands by 18.32% and in the beta and gamma bands by 4.34% with smaller standard deviations. This result confirmed the effectiveness of considering more diverse neurophysiological interactions within and between frequency bands for Riemannian EEG classification.

International audience; We present ComBiNet, a visualization, query, and comparison system for exploring bipartite multivariate dynamic social networks. Historians and sociologists study social networks constructed from textual sources mentioning events related to people, such as marriage acts, birth certificates, and contracts. We model this type of data using bipartite multivariate dynamic networks to maintain a representation faithful to the original sources while not too complex. Relying on this data model, ComBiNet allows exploring networks using both visual and textual queries using the Cypher language, the two being synchronized to specify queries using the most suitable modality; simple queries are easy to express visually and can be refined textually when they become complex. These queries are used for applying topological and attribute-based selection on the network. Query results are visualized in the context of the whole network and over a geographical map for geolocalized entities. We also present the design of our interaction techniques for querying social networks to visually compare the selections in terms of topology, measures, and attribute distributions. We validate the query and comparison systems by showing how they have been used to answer historical questions and by explaining how they have been improved through a usability study conducted with historians.

International audience; Establishing an effective collaboration between augmented-reality (AR) and remote desktop users is a challenge because collaborators do not share a common physical space and equipment. Yet, such asymmetrical collaboration configurations are common today for many design tasks, due to the geographical distance of people or unusual circumstances such as a lockdown. We conducted a first study to investigate trade-offs of three remote representations of an AR workspace: a fully virtual representation, a first-person view, and an external view. Building on our findings, we designed ARgus, a multi-view video-mediated communication system that combines these representations through interactive tools for navigation, previewing, pointing, and annotation. We report on a second user study that observed how 12 participants used ARgus to provide remote instructions for an AR furniture arrangement task. Participants extensively used its view transition tools, while the system reduced their reliance on verbal instructions.

International audience; We contribute a research agenda for visualization in motion and two experiments to understand how well viewers can read data from moving visualizations. We define visualizations in motion as visual data representations that are used in contexts that exhibit relative motion between a viewer and an entire visualization. Sports analytics, video games, wearable devices, or data physicalizations are example contexts that involve different types of relative motion between a viewer and a visualization. To analyze the opportunities and challenges for designing visualization in motion, we show example scenarios and outline a first research agenda. Motivated primarily by the prevalence of and opportunities for visualizations in sports and video games we started to investigate a small aspect of our research agenda: the impact of two important characteristics of motion---speed and trajectory on a stationary viewer's ability to read data from moving donut and bar charts. We found that increasing speed and trajectory complexity did negatively affect the accuracy of reading values from the charts and that bar charts were more negatively impacted. In practice, however, this impact was small: both charts were still read fairly accurately.

We present a case study on a journey about a personal data collection of carnivorous plant species habitats, and the resulting scientific exploration of location data biases, data errors, location hiding, and data plausibility. While initially driven by personal interest, our work led to the analysis and development of various means for visualizing threats to insight from geo-tagged social media data. In the course of this endeavor we analyzed local and global geographic distributions and their inaccuracies. We also contribute Motion Plausibility Profiles-a new means for visualizing how believable a specific contributor's location data is or if it was likely manipulated. We then compared our own repurposed social media dataset with data from a dedicated citizen science project. Compared to biases and errors in the literature on traditional citizen science data, with our visualizations we could also identify some new types or show new aspects for known ones. Moreover, we demonstrate several types of errors and biases for repurposed social media data. Please note that people with color impairments may consider our alternative paper version.

International audience; We present how we co-designed with a connoisseur a system that visualizes a star-like ribbon joining at the solar plexus and animated it from motion capture data to perform Isadora Duncan’s dances. Additionally, the system visualizes the trace of the solar plexus and specific keyframes of the choreography as key poses placed in the 3D space. We display the visualization in a Hololens headset and provide features that allow to manipulate it in order to understand and learn Duncan’s qualities and choreographic style. Through a workshop with dancers, we ran a structured observation where we compared qualitatively how the dancers were able to perceive Duncan’s qualities and embody them using the system set according to two different conditions: displaying all the future keyframes or displaying a limited number of keyframes. We discuss the results of our workshop and the use of augmented reality in the studio for pedagogical purposes.

Discontinuities and delayed terms are encountered in the governing equations of a large class of problems ranging from physics, engineering, medicine to economics. These systems are impossible to be properly modelled and simulated with standard Ordinary Differential Equations (ODE), or any data-driven approximation including Neural Ordinary Differential Equations (NODE). To circumvent this issue, latent variables are typically introduced to solve the dynamics of the system in a higher dimensional space and obtain the solution as a projection to the original space. However, this solution lacks physical interpretability. In contrast, Delay Differential Equations (DDEs) and their data-driven, approximated counterparts naturally appear as good candidates to characterize such complicated systems. In this work we revisit the recently proposed Neural DDE by introducing Neural State-Dependent DDE (SDDDE), a general and flexible framework featuring multiple and state-dependent delays. The developed framework is auto-differentiable and runs efficiently on multiple backends. We show that our method is competitive and outperforms other continuous-class models on a wide variety of delayed dynamical systems.

International audience; Today’s artificial intelligence, largely based on data-intensive machine learning algorithms, relies heavily on the digital labour of invisibilized and precarized humans-in-the-loop who perform multiple functions of data preparation, verification of results, and even impersonation when algorithms fail. Using original quantitative and qualitative data, the present article shows that these workers are highly educated, engage significant (sometimes advanced) skills in their activity, and earnestly learn alongside machines. However, the loop is one in which human workers are at a disadvantage as they experience systematic misrecognition of the value of their competencies and of their contributions to technology, the economy, and ultimately society. This situation hinders negotiations with companies, shifts power away from workers, and challenges the traditional balancing role of the salary institution.

International audience; Existing studies in black-box optimization for machine learning suffer from low generalizability, caused by a typically selective choice of problem instances used for training and testing different optimization algorithms. Among other issues, this practice promotes overfitting and poor-performing user guidelines. To address this shortcoming, we propose in this work a benchmark suite, OptimSuite, which covers a broad range of black-box optimization problems, ranging from academic benchmarks to real-world applications, from discrete over numerical to mixed-integer problems, from small to very large-scale problems, from noisy over dynamic to static problems, etc. We demonstrate the advantages of such a broad collection by deriving from it Automated Black Box Optimizer (ABBO), a general-purpose algorithm selection wizard. Using three different types of algorithm selection techniques, ABBO achieves competitive performance on all benchmark suites. It significantly outperforms previous state of the art on some of them, including YABBOB and LSGO. ABBO relies on many high-quality base components. Its excellent performance is obtained without any task-specific parametrization. The OptimSuite benchmark collection, the ABBO wizard and its base solvers have all been merged into the open-source Nevergrad platform, where they are available for reproducible research.

International audience; We study tangible touch tablets combined with Augmented Reality Head-Mounted Displays (AR-HMDs) to perform spatial 3D selections. We are primarily interested in the exploration of 3D unstructured datasets such as cloud points or volumetric datasets. AR-HMDs immerse users by showing datasets stereoscopically, and tablets provide a set of 2D exploration tools. Because AR-HMDs merge the visualization, interaction, and the users’ physical spaces, users can also use the tablets as tangible objects in their 3D space. Nonetheless, the tablets’ touch displays provide their own visualization and interaction spaces, separated from those of the AR-HMD. This raises several research questions compared to traditional setups. In this paper, we theorize, discuss, and study different available mappings for manual spatial selections using a tangible tablet within an AR-HMD space. We then study the use of this tablet within a 3D AR environment, compared to its use with a 2D external screen.

International audience; Hypergraphs are a generalization of graphs in which edges (hyperedges) can connect more than two vertices-as opposed to ordinary graphs where edges involve only two vertices. Hypergraphs are a fairly common data structure but there is little consensus on how to visualize them. To optimize a hypergraph drawing for readability, we need a layout algorithm. Common graph layout algorithms only consider ordinary graphs and do not take hyperedges into account. We focus on layered hypergraphs, a particular class of hypergraphs that, like layered graphs, assigns every vertex to a layer, and the vertices in a layer are drawn aligned on a linear axis with the axes arranged in parallel. In this paper, we propose a general method to apply layered graph layout algorithms to layered hypergraphs. We introduce six different transformations for layered hypergraphs. The choice of transformation affects the subsequent graph layout algorithm in terms of computational performance and readability of the results. Thus, we perform a comparative evaluation of these transformations in terms of number of crossings, edge length, and impact on performance. We also provide two case studies showing how our transformations can be applied to real-life use cases. A copy of this paper with all appendices and supplemental material is available at osf.io/grvwu.

International audience; Electroencephalography-based brain-computer interface (EEG-BCI) systems have been developed to enable individuals with physical disabilities to control external devices using their thoughts. The performance of EEG-BCI systems can be affected by various factors, including the choice of EEG feature representation.Contrastive learning is a machine learning approach that involves training a model to learn a discriminative representation that captures the underlying patterns in the data. This method has been shown to be effective for learning representations in a variety of domains, including computer vision, natural language processing, and speech recognition. This process encourages the model to learn features that capture the essential information needed to distinguish between different classes while ignoring the irrelevant information.In this study, we propose a contrastive learning approach to learn discriminative representations of EEG signals in the context of MI tasks. Specifically, we investigate the use of different EEG signal representations based on symmetric positive definite (SPD) matrix: covariance matrix, augmented covariance matrix, and coherence matrix. To achieve this, we process the different representation using the SPDNet architecture, which is designed to process SPD matrices. The main advantage of contrastive learning is that it can learn representations that are invariant to irrelevant variations, such as differences in electrode placement, noise, and individual differences in brain activity. By learning to contrast similar and dissimilar examples, contrastive learning can discover patterns that are shared across subjects and generalize well to new data.We validate our approach using the MOABB package, focusing on Cross Subject evaluation. In this sense contrastive learning can be used to learn meaningful representations of EEG signals from different subjects. The study is in its initial stages and further research is required to fully explore the potential of contrastive learning and SPDNet in the context of EEG-BCI.

Workshop, Due to the COVID-19 pandemic, the event has been cancelled.; International audience; People can generate more innovative ideas when they collaborate with one another, collectively exploring ideas and exchanging viewpoints. Advancements in artificial intelligence have opened up new opportunities in people’s creative activities where individual users ideate with diverse forms of AI. For instance, AI agents and intelligent tools have been designed as ideation partners that provide inspiration, suggest ideation methods, or generate alternative ideas. However, what AI can bring to collaborative ideation among a group of users has not been fully understood. Compared to ideating with individuals, ideating with multiple users would require understanding users’ social interaction, transforming individual efforts into a group effort, and—in the end—making users satisfied that they collaborated with other group members. This workshop aims to bring together a community of researchers and practitioners to explore the integration of AI in human-human collaborative ideation. The exploration will center around identifying the potential roles of AI as well as the process and form of collaborative ideation, considering what users want to do with AI or humans.

International audience; User interfaces typically feature tools to act on objects and rely on the ability of users to discover or learn how to interact with them. Previous work in HCI has used the Theory of Affordances to explain how users understand the possibilities for action in digital environments. A complementary theory from cognitive neuroscience, Technical Reasoning, posits that users accumulate abstract knowledge of object properties and technical principles known as mechanical knowledge, essential in tool use. Drawing from this theory, we introduce interaction knowledge as the "mechanical" knowledge of digital environments. We provide evidence of its relevance by reporting on an experiment where participants performed tasks in a digital environment with ambiguous possibilities for interaction. We analyze how interaction knowledge was transferred across two digital domains, text editing and graphical editing, and conclude that interaction knowledge models an essential type of knowledge for interacting in the digital world. CCS CONCEPTS • Human-centered computing → HCI theory, concepts and models.

International audience; Wall displays are well suited for collaborative work and are often placed in rooms with ample space in front of them that remains largely unused. Augmented Reality (AR) headsets can seamlessly extend the collaboration space around the Wall. Nevertheless, it is unclear if extending Walls with AR is effective and how it may affect collaboration. We first present a prototype combining a Wall and AR headsets to extend the Wall workspace. We then use this prototype to study how users utilize the virtual space created in AR. In an experiment with 24 participants, we compare how pairs solve collaborative tasks with the Wall alone and with Wall+AR. Our qualitative and quantitative results highlight that with Wall+AR, participants use the physical space in front and around the Wall extensively, and while this creates interaction overhead, it does not impact performance and improves the user experience.

International audience; Large vertical surfaces such as wall displays allow users to work with a very large and high-resolution workspace. Such displays promote physical navigation: users can step close to the display to see details, but also move away to get a wider view of the workspace. In terms of input, current solutions usually combine direct touch on the wall with input on a handheld device, disconnecting close and distant input rather than treating it as a continuum. We present SurfAirs, which are physical controllers that users can manipulate on screen (surface input), in the air (mid-air input), and transition from the surface to the air during a single manipulation (hybrid input). We report on two user studies that compare SurfAirs' performance with bare-hand input for both mid-air and hybrid input. Participants prefer and perform better with SurfAirs.

International audience; We present three studies that investigate the effectiveness of multiple glanceable part-to-whole proportion representations on smartwatch faces. Our goal was to understand how quickly and accurately people can make judgments about their progress toward multiple goals displayed in a small space. We designed our three studies with increasing external validity. The first study compared bar charts, radial bar charts, and text representations-shown with a digital time display. The second study added an analog time dial as a distractor to increase the complexity of the watch face. To emulate realistic viewing conditions, the third study investigated the effect of viewing angles. In Study 1 bar and radial bar charts outperformed text representations, in Study 2 adding an analog time dial as a distractor did not affect task performance, and in Study 3 only the most extreme angle led to some performance decrease. Supplementary material is available at https://osf.io/ad2z7/.

Wall displays immerse users in large, high-resolution information spaces. They are well suited for data analysis, as users only need to move around the physical space to explore the virtual information space displayed on the wall. They also facilitate collaboration as their large physical size can accommodate multiple users. However, designing effective ways of interacting with wall displays is challenging. Traditional input devices, such as mice and keyboards, quickly show their limitations in an environment where multiple users can interact and move freely.HCI literature offers interesting alternatives to traditional input techniques. In particular, Tangible User Interactions (TUIs), where users rely on custom tangible objects to interact with the virtual scene, have proved efficient with different types of displays ranging from smartphones to tabletops. Tangible controllers have natural advantages such as the haptic feedback they provide that enables eyes-free manipulations. They also afford specific grasps and manipulations, guiding users on what they can do with them. Empirical studies that compare tangibles to other forms of input also report quantitative gains in regarding manipulation speed and precision in different hardware setups.However, designing tangible controllers for wall displays is difficult. First, the large size and vertical orientation of walls must be taken into account to design tangibles with a suitable form factor. Second, users move in space. They move away to get a wider view, move closer to see details, or adjust their physical position based on other users and objects in the room. This means that tangible controllers must be usable regardless of the user's position in the room, which has some impact on design and engineering aspects. Finally, a wall display is often located in an environment that feature other devices and displays. In such cases, designing tangible controllers for a wall display requires to consider the whole multi-display environment, which constrains even more the tangibles' form factor and the underlying technologies.My thesis work makes three contributions towards enabling tangible interaction with wall displays.The first project, WallTokens, contributes tangibles for enabling on surface interaction with wall displays. WallTokens are low-cost, passive controllers that users can manipulate directly on the wall's surface. WallTokens have a mechanism that allows users to easily attach and detach them from the wall surface, so that when users are done interacting, they can leave them in place and free their hands for other purposes. We report on two studies assessing WallTokens' usability, showing that they are more precise and comfortable than bare-hand gestures to perform low-level manipulations on walls.The second project, SurfAirs, contributes tangibles that support not only on surface interaction but also distant interaction with wall displays. We present two possible designs for versatile tangible controllers that can be used both on the wall surface when users need precision and detail, and in the air when they need a wide viewing angle. SurfAirs support both types of input, as well as smooth transitions between the two. We report on two studies that compare SurfAir prototypes with bare hand gestures for performing low-level manipulation tasks. SurfAirs outperform bare hand gestures regarding accuracy, speed and user preference.The third project contributes a survey about the use of physical controllers to interact with a physical display. Each project is described along twelve dimensions that capture the design aspects of the controller, the properties of the display and how they communicate with each other. We contribute a Web page to explore this list of references along the different dimensions, and use it to discuss the challenges that underlie the design of tangible controllers in a multi-display environment.; Les murs d'écrans plongent les utilisateurs dans de larges espaces d'informations ultra-haute résolution. Ils sont bien adaptés à l'analyse de grands ensembles de données car les utilisateurs peuvent se déplacer physiquement pour explorer ce qui est affiché à l'écran. Ils facilitent également la collaboration car leur taille permet facilement d'accueillir plusieurs utilisateurs à la fois. Cependant, créer des interactions efficaces avec les murs d'écrans est un défi. Les périphériques traditionnels tel que le clavier et la souris montre vite leur limite dans un contexte ou plusieurs utilisateurs interagissent et se déplacent librement.La littérature en IHM propose des moyens d'interaction alternatifs. En particulier, l'interaction tangible, qui s'appuie sur la manipulation d'objets physiques pour interagir avec des scènes virtuelles, offre des avantages intéressants pour les murs d'écrans. La matérialité des contrôleurs offre un retour haptique qui permet de les manipuler sans les regarder. Leur forme suggère aussi la manière de les attraper et de les manipuler, guidant les utilisateurs sur leurs fonctionnalités. De nombreuses études empiriques comparant les contrôleurs tangibles à d'autres formes d'interaction montrent qu'ils procurent des gains significatifs en termes de vitesse et de précision de manipulation.Cependant, fabriquer des contrôleurs tangibles pour les murs d'écrans est difficile. Tout d'abord, l'orientation de l'écran et sa taille doivent être prise en compte pour fabriquer des tangibles appropriés. De plus, les utilisateurs sont mobiles : ils s'éloignent pour obtenir un angle de vue plus large, se rapprochent pour voir plus de détails, ou ajustent leur position en fonction de celle des autres utilisateurs. Les contrôleurs tangibles doivent donc être conçu pour être utilisable quelle que soit la position de l'utilisateur dans la pièce. Enfin, un mur d'écrans est souvent situé dans un contexte rassemblant d'autres dispositifs (tables interactives, ordinateurs, etc.). Dans ce cas, il est nécessaire de prendre en compte l'ensemble du contexte, contraignant la forme des tangibles et les technologies sous-jacentes.Mon travail de thèse propose trois contributions pour faciliter l'interaction tangible avec les murs d'écrans.Mon premier projet, WallTokens, propose des tangibles qui permettent d'interagir sur la surface des murs d'écrans. Les WallTokens sont équipés d'un mécanisme qui permet aux utilisateurs de les attacher et de les détacher facilement de la surface du mur. Cela permet de les laisser en place lorsque les utilisateurs veulent libérer leur main pour d'autres tâches. Nous présentons deux études évaluant la facilité d'utilisation et l'efficacité des WallTokens. Nos résultats montrent qu'ils sont plus précis et plus confortables que les interactions tactiles pour effectuer des manipulations de bas niveau sur mur d'écrans.Mon deuxième projet, SurfAirs, propose des tangibles permettant des interactions avec les murs d'écrans en surface, quand les utilisateurs ont besoin de détails et de précision, mais aussi à distance quand ils ont besoin d'un grand angle de vue. Les SurfAirs permettent également une transition continue entre ces deux modes d'interaction. Nous présentons deux études qui comparent les SurfAirs avec des gestes à main nue pour effectuer des tâches de manipulation de bas niveau. Les SurfAirs sont plus performants que les gestes à main nue en termes de précision et de vitesse et les utilisateurs les préfèrent.Le troisième projet propose une étude de la littérature sur l'utilisation de contrôleur tangible avec des écrans. Chaque article étudié est classifié selon 12 dimensions qui reflètent les aspects de la conception du contrôleur et de l'écran. Nous proposons un outil Web qui permet l'exploration de notre corpus d'articles à travers ces dimensions de classification. Nous discutons ensuite les défis qui sous-tendent la conception de contrôleurs tangibles dans un environnement multi-écrans.

International audience; We investigate graph transformations, defined using Datalog-like rules based on acyclic conjunctive two-way regular path queries (acyclic C2RPQs), and we study two fundamental static analysis problems: type checking and equivalence of transformations in the presence of graph schemas. Additionally, we investigate the problem of target schema elicitation, which aims to construct a schema that closely captures all outputs of a transformation over graphs conforming to the input schema. We show all these problems are in EXPTIME by reducing them to C2RPQ containment modulo schema; we also provide matching lower bounds. We use cycle reversing to reduce query containment to the problem of unrestricted (finite or infinite) satisfiability of C2RPQs modulo a theory expressed in a description logic.

International audience; Marcelle is an open-source toolkit dedicated to the design of web applications involving human interactions with machine learning algorithms. This demonstration illustrates how Marcelle can support research and education in the field of interactive machine learning, notably in scenarios involving multiple stakeholders.; Marcelle est un toolkit open-source dédiée à la conception d’applications web impliquant des interac- tions avec des algorithmes d’apprentissage machine. Cette démonstration illustre comment Marcelle peut stimuler la recherche et l’éducation dans le domaine de l’apprentissage automatique interactif, notamment dans des scénarios impliquant plusieurs types d’utilisateurs.

International audience; Previous literature on small-scale shape displays demonstrated the benefit of shape-changing interfaces and adaptive environments. However, research with large-scale implementations is limited to specific equipment. Thus, there is little understanding of how they can enrich user experience in a larger scope, especially from an interactive perspective. We proposed using pin-array large-scale implementations made of linear actuators matrix as vertical structures to replace static and planar walls and conducted a two-fold exploration to extend the understanding of potential applications and showcase users’ behaviors and preferences to inspire researchers. We summarized 14 potential applications for shape-changing walls, and presented users’ interaction preferences with such a wall of different sizes and interaction distances. We concluded with recommendations and guidelines for practitioners to design large-scale actuated walls.

My habilitation narrates my past ten years of dance-led research in Human-Computer Interaction and Interaction Design, where dance practice served as the experimental frame through which I designed and integrated interactive systems. I address two main questions: How does the integration of technologies in dance operate?What types of opportunities, understandings, tensions, and discourses emerge from such integration?The habilitation emphasizes my epistemological journey from quantitative methodologies to research through practice to research-creation. These distinct methodologies allowed me to design a series of interactive systems spanning from generic ones for dance movement analysis to systems grounded in dance artists’ practices and systems designed for my own personal choreographic work. This journey also involved progressively letting my personal subjective voice come through in my academic work.; Mon habilitation retrace mes dix dernières années de recherche par la danse en Interaction Humain-Machine et en Design d’interaction, où la pratique de la danse agit comme un cadre expérimental dans lequel je conçois et intègre des systèmes interactifs. J’y réponds à deux questions principales :Comment s'opère l'intégration des technologies en danse ?Quels types d'opportunités, de tensions et de discours émergent d'une telle intégration ?L'habilitation met l'accent sur mon parcours épistémologique allant de méthodologies quantitatives à la recherche par la pratique et la recherche-création. Ces méthodologies m'ont permis de concevoir une série de systèmes interactifs allant de systèmes génériques pour l'analyse du mouvement dansé à des systèmes centrés sur les pratiques des danseurs et des systèmes conçus pour mon propre travail chorégraphique. Ce parcours a également consisté à laisser progressivement transparaître ma voix personnelle dans mon travail académique.

This thesis covers research on how to design and use micro-visualizations for pervasive and mobile data exploration on smartwatches and fitness trackers. People increasingly wear smartwatches that can track and show a wide variety of data. My work is motivated by the potential benefits of data visualizations on small mobile devices such as fitness monitoring armbands and smartwatches. I focus on situations in which visualizations support dedicated data-related tasks on interactive smartwatches. My main research goal in this space is to understand more broadly how to design small-scale visualizations for fitness trackers. Here, I explore: (i) design constraints in the small space through an ideation workshop; (ii) what kind of visualizations people currently see on their watch faces; (iii) a design review and design space of small-scale visualizations; (iv) and readability of micro-visualizations considering the impact of size and aspect ratio in the context of sleep tracking. The main findings of the thesis are, first, a set of data needs concerning a sightseeing usage context in which these data needs were met with a wealth of dedicated visualization designs that go beyond those commonly seen on watch displays. Second, a predominant display of health & fitness data, with icons accompanying the text being the most frequent representation type on current smartwatch faces. Third, a design space for smartwatch face visualizations which highlights the most important considerations for new data displays for smartwatch faces and other small displays. Last, in the context of sleep tracking, we saw that people performed simple tasks effectively, even with complex visualization, on both smartwatch and fitness band displays; but more complex tasks benefited from the larger smartwatch size. Finally, I point out important open opportunities for future smartwatch visualization research, such as scalability (e.g., more data, smaller size, and more visualizations), the role of context and wearer's movement, smartwatch display types, and interactivity. In summary, this thesis contributes to the understanding of visualizations on smartwatches and highlights open opportunities for smartwatch visualization research.; Cette thèse porte sur la recherche sur la conception et l'utilisation de micro-visualisations pour l'exploration de données mobiles et pervasives sur des smartwatches et des trackers de fitness. Les gens portent de plus en plus de smartwatches qui peuvent suivre et afficher une grande variété de données. Mon travail est motivé par les avantages potentiels des visualisations de données sur les petits appareils mobiles tels que les brassards de suivi de la condition physique et les smartwatches. Je me concentre sur les situations dans lesquelles les visualisations soutiennent des tâches spécifiques liées aux données sur des smartwatches interactives. Mon principal objectif de recherche dans ce domaine est de comprendre plus largement comment concevoir des visualisations à petite échelle pour les trackers de fitness. Ici, j'explore : (i) les contraintes de conception dans le petit espace par le biais d'un atelier d'idéation ; (ii) le type de visualisations que les gens voient actuellement sur le visage de leur montre ; (iii) une revue de conception et l'espace de conception des visualisations à petite échelle ; (iv) et la lisibilité des micro-visualisations en considérant l'impact de la taille et du rapport d'aspect dans le contexte du suivi du sommeil. Les principaux résultats de la thèse sont, premièrement, un ensemble de besoins de données concernant un contexte d'utilisation touristique dans lequel ces besoins de données ont été satisfaits avec une richesse de conceptions de visualisation dédiées qui vont au-delà de celles couramment vues sur les affichages des montres. Deuxièmement, un affichage prédominant des données de santé et de forme physique, les icônes accompagnant le texte étant le type de représentation le plus fréquent sur les faces actuelles des smartwatchs. Troisièmement, un espace de conception pour les visualisations sur les faces de smartwatch qui met en évidence les considérations les plus importantes pour les nouveaux affichages de données sur les faces de smartwatch et autres petits écrans. Enfin, dans le contexte du suivi du sommeil, nous avons constaté que les gens effectuaient des tâches simples de manière efficace, même avec une visualisation complexe, à la fois sur les écrans de la smartwatch et du bracelet de fitness, mais que les tâches plus complexes bénéficiaient de la taille plus grande de la smartwatch. Dans la thèse, je souligne les opportunités ouvertes importantes pour les futures recherches sur la visualisation des smartwatchs, telles que l'évolutivité (par exemple, plus de données, une taille plus petite et plus de visualisations), le rôle du contexte et du mouvement du porteur, les types d'affichage des smartwatchs et l'interactivité. En résumé, cette thèse contribue à la compréhension des visualisations sur les smartwatches et met en évidence les opportunités ouvertes pour la recherche en visualisation sur les smartwatches.

IHM'23 - 34e Conférence Internationale Francophone sur l'Interaction Humain-MachineAFIHM, Université de Technologie de Troyes; In Augmented Reality environments, multiple designers can interact with virtual content that overlays their shared physical objects. However, conflicts arise when two designers try to add or modify content around the same virtual space, for example, to explore a new path of ideas. Likewise, although content created by one designer can serve as a source of inspiration for others, it can also distract their creative process. To address those problems, we introduce a conceptual framework that allows multiple versions of augmentations of the same physical object to coexist in parallel virtual spaces. According to our framework, users can partially or totally desynchronize their virtual environment to generate their own content and then explore alternatives created by others. To illustrate it, we present a scenario of collaborative design in Augmented Reality, where two designers sketch in 3D around a physical sewing mannequin to design female dress.; La Réalité Augmentée permet de faciliter la conception collaborative en superposant du contenu virtuel sur un objet physique. Cependant, des conflits peuvent survenir dans le partage de l’espace 3D lorsque plusieurs concepteurs ajoutent ou modifient du contenu virtuel au même endroit pour développer des idées différentes. De plus, même si la visibilité du contenu créé par un concepteur peut renforcer la créativité de ses collaborateurs, elle peut également être source d’inhibition. Pour répondre à ces problèmes, nous introduisons un framework conceptuel permettant à plusieurs versions des augmentations d’un même objet physique de coexister dans des espaces virtuels parallèles. Les utilisateurs peuvent ainsi partiellement ou totalement désynchroniser leur espace virtuel afin de générer leur propre contenu, puis explorer les alternatives créées par leur collaborateur. Pour illustrer notre concept, nous présentons un scénario de conception collaborative d’un vêtement sur un mannequin de couture physique grâce à du dessin 3D en Réalité Augmentée.

IHM 2023 - 34e Conférence Internationale Francophone sur l'Interaction Humain-MachineAFIHM, Université de Technologie de Troyes; This paper introduces a new interaction model based on the concepts of interaction substrate for organizing digital information and interaction instruments for manipulating these substrates. This approach makes the concept of application unnecessary. Instead, it leads to flexible and extensible environments in which users can combine content at will and choose the tools they need to manipulate it. We present Stratify, a proof-of-concept implementation that combines a data-reactive approach to specify relationships among digital objects with a functional-reactive approach to handle interaction. This combination enables the creation of rich information substrates that can be freely inspected and modified, as well as interaction instruments that are decoupled from the objects they interact with, making it possible to use instruments with objects they were not designed for. We illustrate the flexibility of the approach with several examples and present directions for future work.; Cet article présente un nouveau modèle d’interaction fondé sur les concepts de substrat d’interaction pour organiser l’information numérique et d’instruments d’interaction pour manipuler ces substrats. Cette approche permet de s’affranchir du concept d’application et de créer des environnements flexibles et extensibles dans lesquels les utilisateurs peuvent combiner à loisir les contenus et choisir les instruments qui leur conviennent pour les manipuler. Nous présentons Stratify, une implémentation de ces concepts qui combine une approche réactive aux données pour spécifier les relations entre les objets numériques et une approche fonctionnelle réactive pour gérer l’interaction. Cette combinaison permet de créer des substrats d’information riches qui peuvent être librement inspectés et modifiés, ainsi que des instruments d’interaction qui sont découplés des objets avec lesquels ils interagissent, ce qui permet d’utiliser les instruments avec des objets pour lesquels ils n’ont pas été conçus. Nous illustrons la flexibilité de l’approche à l’aide d’exemples et présentons des pistes de développements futurs.

With the increasing interest in movement sonification and expressive gesture-based interaction, it is important to understand which factors contribute to movement learning and how. We explore the effects of movement sonification and users’ musical background on motor variability in complex gesture learning. We contribute an empirical study in which musicians and non-musicians learn two gesture sequences over three days, with and without movement sonification. Results show the interlaced interaction effects of these factors and how they unfold in the three-day learning process. For gesture 1, which is fast and dynamic with a direct “action-sound” sonification, movement sonification induces higher variability for both musicians and non-musicians on day 1. While musicians reduce this variability to a similar level as no auditory feedback condition on day 2 and day 3, non-musicians remain to have significantly higher variability. Across three days, musicians also have significantly lower variability than non-musicians. For gesture 2, which is slow and smooth with an “action-music” metaphor, there are virtually no effects. Based on these findings, we recommend future studies to take into account participants’ musical background, consider longitudinal study to examine these effects on complex gestures, and use awareness when interpreting the results given a specific design of gesture and sound.

International audience; We present a visual analytics tool, MiningVis, to explore the long-term historical evolution and dynamics of the Bitcoin mining ecosystem. Bitcoin is a cryptocurrency that attracts much attention but remains difficult to understand. Particularly important to the success, stability, and security of Bitcoin is a component of the system called "mining.'' Miners are responsible for validating transactions and are incentivized to participate by the promise of a monetary reward. Mining pools have emerged as collectives of miners that ensure a more stable and predictable income. MiningVis aims to help analysts understand the evolution and dynamics of the Bitcoin mining ecosystem, including mining market statistics, multi-measure mining pool rankings, and pool hopping behavior. Each of these features can be compared to external data concerning pool characteristics and Bitcoin news. In order to assess the value of MiningVis, we conducted online interviews and insight-based user studies with Bitcoin miners. We describe research questions tackled and insights made by our participants and illustrate practical implications for visual analytics systems for Bitcoin mining.

International audience; We present V-Mail, a framework of cross-platform applications, interactive techniques, and communication protocols for improved multi-person correspondence about spatial 3D datasets. Inspired by the daily use of e-mail, V-Mail seeks to enable a similar style of rapid, multi-person communication accessible on any device; however, it aims to do this in the new context of spatial 3D communication, where limited access to 3D graphics hardware typically prevents such communication. The approach integrates visual data storytelling with data exploration, spatial annotations, and animated transitions. V-Mail "data stories" are exported in a standard video file format to establish a common baseline level of access on (almost) any device. The V-Mail framework also includes a series of complementary client applications and plugins that enable different degrees of story co-authoring and data exploration, adjusted automatically to match the capabilities of various devices. A lightweight, phone-based V-Mail app makes it possible to annotate data by adding captions to the video. These spatial annotations are then immediately accessible to team members running high-end 3D graphics visualization systems that also include a V-Mail client, implemented as a plugin. Results and evaluation from applying V-Mail to assist communication within an interdisciplinary science team studying Antarctic ice sheets confirm the utility of the asynchronous, cross-platform collaborative framework while also highlighting some current limitations and opportunities for future work.

International audience; We developed and validated a rating scale to assess the aesthetic pleasure (or beauty) of a visual data representation: the BeauVis scale. With our work we offer researchers and practitioners a simple instrument to compare the visual appearance of different visualizations, unrelated to data or context of use. Our rating scale can, for example, be used to accompany results from controlled experiments or be used as informative data points during in-depth qualitative studies. Given the lack of an aesthetic pleasure scale dedicated to visualizations, researchers have mostly chosen their own terms to study or compare the aesthetic pleasure of visualizations. Yet, many terms are possible and currently no clear guidance on their effectiveness regarding the judgment of aesthetic pleasure exists. To solve this problem, we engaged in a multi-step research process to develop the first validated rating scale specifically for judging the aesthetic pleasure of a visualization (osf.io/fxs76). Our final BeauVis scale consists of five items, "enjoyable," "likable," "pleasing," "nice," and "appealing." Beyond this scale itself, we contribute (a) a systematic review of the terms used in past research to capture aesthetics, (b) an investigation with visualization experts who suggested terms to use for judging the aesthetic pleasure of a visualization, and (c) a confirmatory survey in which we used our terms to study the aesthetic pleasure of a set of 3 visualizations.

Visualization plays a crucial role in molecular and structural biology. It has been successfully applied to a variety of tasks, including structural analysis and interactive drug design. While some of the challenges in this area can be overcome with more advanced visualization and interaction techniques, others are challenging primarily due to the limitations of the hardware devices used to interact with the visualized content. Consequently, visualization researchers are increasingly trying to take advantage of new technologies to facilitate the work of domain scientists. Some typical problems associated with classic 2D interfaces, such as regular desktop computers, are a lack of natural spatial understanding and interaction, and a limited field of view. These problems could be solved by immersive virtual environments and corresponding hardware, such as virtual reality head-mounted displays. Thus, researchers are investigating the potential of immersive virtual environments in the field of molecular visualization. There is already a body of work ranging from educational approaches to protein visualization to applications for collaborative drug design. This review focuses on molecular visualization in immersive virtual environments as a whole, aiming to cover this area comprehensively. We divide the existing papers into different groups based on their application areas, and types of tasks performed. Furthermore, we also include a list of available software tools. We conclude the report with a discussion of potential future research on molecular visualization in immersive environments. This work was supported by the US DOE LLNL-LDRD 22-FS- 026 and was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under Contract DE-AC52-07NA27344 (LLNL-CONF-831245) and by project TIN2017-88515-C2-1-R (GEN3DLIVE), from the Spanish Ministerio de Economía y Competitividad, by 839 FEDER (EU) funds. The work received further support by the French National Research agency (Grant “PIRATE”, ANR-21-CE45-0014) and by the “Initiative d’Excellence” program from the French State (Grant “DYNAMO”, ANR-11-LABX-0011 and grant “CACSICE”, ANR-11-EQPX-0008), and from Carl Zeiss Stiftung. Ministerio de Ciencia e Innovación/AEI (PID2021-122136OB- C21[10.13039/501100011033/FEDER, UE])

International audience; Participatory design actively involves users throughout the design process, from initial discovery of their needs to the final assessment of the system. As in all forms of interaction design, this requires two complementary processes: generating new ideas that expand the design space, and selecting specific ideas, thus contracting the design space. Interaction designers create prototypes that help them generate and explore this space of possibilities, ideally with the ongoing participation of users. Successful prototypes represent different aspects of the system that highlight specific design questions. We classify prototypes based on: Representation, which refers to their physical form, ranging from paper, cardboard or foam mock-ups to complete software or hardware simulations; Precision, which refers to the level of detail, ranging from simple hand-drawn sketches or physical mock-ups to polished computer-generated images or 3D printed objects; Interactivity, which refers to the level of interaction possible, ranging from no interaction, such as a video scenario, to partial interaction, when a designer "walks through" a scenario, to fully interactive, either simulating interaction with the Wizard-of-Oz method or trying an operational software or hardware prototype; Life cycle, which refers to the expected evolution of the prototype, from "throw away" prototypes in the earliest design phase to successively more developed prototypes in later design phases, to actual components of the final system; and Scope, which refers to the aspects of the final system covered by the prototype, including breadth-first horizontal prototypes, depth-first vertical prototypes, or path-based story prototypes. This chapter explains how to create and use these different kinds of prototypes, with examples selected from key phases of the interaction design process.

The amount of data produced every day is constantly growing. This mass of raw information requires finding solutions to analyze and visualize it.Wall displays are a good candidate to solve this issue, allowing to display a large amount of data.However, they are shared devices made to accommodate several people, which does not allow for the existence of a private workspace. Moreover, wall displays are still physical devices that can not be easily extended when the content to display becomes too large.In this manuscript, I explore the combination of Augmented Reality (AR) and a wall display to: first, add a personalized information space onto a wall display; and second, extend the rendering space. We focus on using an augmented reality headset (AR-HMD) to display content complementary to the context offered by the wall. We study the synergy of these two devices through two research questions: RQ1 - How can AR assist exploration and navigation of a network on a wall display, through personal view and navigation?RQ2 - How does extending the screen space of a wall display using AR changes users’ space exploitation during a collaborative session?For RQ1, I study the case of using AR with a wall display for navigating networks. I present four techniques, proposing different visual aid mechanisms for navigation. I then evaluate these four techniques in two experiments, one for low accuracy path following in the network (path selection) and one for high accuracy path following (path tracing). We report that for the path selection task, a persistent connection between the cursor in AR and the network on the wall obtained better results. In the case of path tracing, we observe that a lighter connection offered better performance. Moreover, we show the feasibility of a system where the interaction is done privately in AR for network navigation on shared screens.For RQ2, I focus on the space available in front of the wall display and consider using AR to extend the wall display. I present a system extending a wall display with AR, taking advantage of the space in front of the wall. We compare our system that combines a wall display with AR with a wall display alone using two collaborative tasks.We observe that users extensively use the available virtual space with our system. Although this creates an additional cost of interaction, we observe no performance difference and a real benefit of this extra space.The complexity of setting up the previously studied systems led us to study a cheaper way to use a wall display: RQ3 - Can we emulate a wall display inside a VR headset?This question has many aspects, and I focus on the capacity for VR headsets to reach the necessary resolution to replicate the user experience of a high-resolution wall display. For this, we study the optic model of VR headsets and compare it to the human vision model. Our analysis indicates that current headsets need a higher resolution to emulate a wall display. We confirm our analysis by running a pilot study comparing a wall display, two VR headsets, and an emulation of a perfect VR headset.I conclude this manuscript by discussing the different ways for a wall display to be combined with AR headsets, the requirements for replacing wall displays by VR/AR headsets, and by elabo- rating on the future work this thesis opens.; La quantité de données produite chaque jour est en croissance constante. Cette masse d’information demande de trouver des solutions pour les analyser et les visualiser. Les murs d’écrans sont de bon candidat pour résoudre ce problème, car ils permettent d’afficher une grande quantité de données. Cependant, ce sont des écrans partagés, conçus pour plusieurs personnes, ce qui ne permet pas d’avoir des espaces de travail privés. De plus, un mur d’écran reste un périphérique d’affichage physique qui est coûteux à agrandir lorsque la quantité de contenu à afficher devient trop grande.Dans ce manuscrit, j’explore la combinaison de la Réalité Augmentée (RA) et d’un mur d’écrans, pour premièrement, ajouter un affichage d’informations personnalisées; et deuxièmement, étendre virtuellement l’espace de rendu. Nous concentrons nos efforts sur l’étude de la synergie de ces deux dispositifs à travers deux questions de recherche :QR1 - Comment la RA peut-elle aider à l’exploration et la navigation d’un réseau sur un mur d’écrans, en ayant un visuel et une navigation personnelle ?QR2 - Comment l’extension du mur d’écran avec la RA change-t-elle l’exploitation de l’espace par les utilisateurs durant une session de collaboration ?Pour RQ1, j’étudie le cas de l’utilisation de la RA avec un mur d’écrans pour la navigation de réseaux. Nous présentons 4 techniques, proposant différents mécanismes d’aide visuelle à la navigation. Nous évaluons ensuite ces 4 techniques lors de deux expériences, une de parcours de chemin de faible précision (sélection de chemin) et une autre de parcours de chemin de haute précision (tracé de chemin). Pour la tâche de sélection de chemin, une connexion persistante entre le curseur en RA et le réseau sur le mur permet de meilleur résultat.Dans le cas du tracé de chemin, nous observons qu’une connexion plus légère offre une meilleure performance. De plus, nous montrons la viabilité d’un système où l’interaction se fait en RA, pour la navigation de réseau sur des écrans partagés.Pour RQ2, je porte mon attention sur l’espace disponible devant le mur d’écrans, et envisage l’utilisation de cet espace pour étendre le mur d’écran avec de la RA. Je présente un système, étendant le mur d’écrans grâce à la RA en utilisant les dimensions de l’espace disponible devant le Mur. Nous comparons notre système avec un mur d’écran seul avec deux tâches collaboratives.Nous observons qu’avec notre système, les utilisateurs utilisent principalement l’espace virtuel. Bien que cela crée un surcoût pour l’interaction, nous n’observons aucune différence de performance avec le mur d’écran seul, et nous observons un vrai bénéfice à cet espace supplémentaire.La complexité de mise en place des systèmes précédemment étudiés nous a amené à étudier une manière moins coûteuse d’utiliser un mur d’écran: QR3 - Peut-on émuler un mur d’écran avec un casque de RV ?Cette question ayant plusieurs facettes, je me concentre sur la capacité des casques de RV d’atteindre la résolution nécessaire à répliquer l’expérience d’utilisation d’un mur haute résolution. Pour ce faire, nous étudions le modèle optique des casques de RV et le comparons au modèle de la vision humaine. Notre analyse indique que les casques actuels n’ont pas une résolution suffisante pour émuler un mur de haute résolution.Nous confirmons cette analyse avec une experience pilote qui compare un mur d’écrans, deux casques de RV et une émulation d’un casque de RV parfait.Je conclus ce manuscrit en discutant plus largement des différentes manières pour les murs d’écran d’être complétés par les casques d’AR et des implications de leur remplacement par des casques de VR/AR, et j’élabore sur les directions de recherches qu’ouvre cette thèse.

As computer graphics technologies develop, spatial data can be better visualized in the 3D environment so that viewers can observe 3D shapes and positions clearly. Meanwhile, 2D abs- tract visualizations can present summarized information, visualize additional data, and control the 3D view. Combining these two parts in one interface can assist people in finishing complicated tasks, especially in scientific domains, though there is a lack of design guidelines for the interaction. Generally, experts need to analyze large scientific data to finish challenging tasks. For example, in the biological field, biologists need to build the hierarchy tree for an embryo with more than 200 cells. In this case, manual work can be time- consuming and tedious, and machine learning algorithms have the potential to alleviate some of the tedious manual processes to serve as the basis for experts. These predictions, however, contain hierarchical and multi-layer information, and it is essential to visualize them sequentially and progressively so that experts can control their viewing pace and validation. Also, 3D and 2D representations, together with machine learning predictions, need to be visually and interactively connected in the system.In this thesis, we worked on the cell lineage problem for plant embryos as an example to investigate a visualization system and its interaction design that makes use of combinations of 3D and 2D representations as well as visualizations for ma- chine learning. We first investigated the 3D selection interaction techniques for the plant embryo.The cells in a plant embryo are tightly packed together, without any space in between. Traditionaltechniques can hardly deal with such an occlusion problem. We conducted a study to evaluate three different selection techniques, and found out that the combination of the Explosion Selection technique and the List Selection technique works well for people to get access and observe plant cells in an embryo. These techniques can also be extended to other similar densely packed 3D data. Second, we explored the visualization and interaction de- sign to combine the 3D visualizations of a plant embryo with its associated 2D hierarchy tree. We designed a system with such combinations for biologists to examine the plant cells and record the development history in the hierarchy tree. We support the hierarchy building in two directions, both constructing the history top-down using the lasso selection in 3D environment and bottom-up as the traditional workflow does in the hierarchy tree. We also added a neural network model to give predictions about the assignments for biologists to start with. We conducted an evaluation with biologists, which showed that both 3D and 2D representations help with making decisions, and the tool can inspire insights for them. One main drawback was that the performance of the machine learning model was not ideal. Thus, to assist the process and enhance the model performance, in an improved version of our system, we trained five different ML models and visualized the predictions and their associated uncertainty. We performed a study, and the results indicated that our designed ML representations are easy to understand, and that the new tool can effectively improve the efficiency of assigning the cell lineage.; Avec le développement des technologies d’infographie, les données spatiales peuvent être mieux visualisées dans leur environnement 3D afin que les spectateurs puissent observer clairement les formes et les positions 3D. Parallèlement, les visualisations abstraites en 2D peuvent présenter des informations résumées, visualiser des données supplémentaires et contrôler la vue 3D. La combinaison de ces deux représentations en une seule interface peut aider les utilisateurs à entreprendre des tâches complexes, en particulier dans les domaines scientifiques, bien qu’il y ait un manque de directives générales de conception pour l’interaction. En général, les experts doivent analyser de volumineuses données scientifiques pour mener à bien des tâches difficiles. Par exemple, dans le domaine biologique, les biologistes doivent construire l’arbre de lignage cellulaire d’un embryon contenant plus de 200 cellules. Dans ce cas, le travail manuel peut être long et fastidieux, et les algorithmes d’apprentissage automatique ont le potentiel d’alléger certains des processus manuels fastidieux en fournissant des annotations ou prédictions initiales aux experts. Dans le cas du lignage cellulaire, ces prédictions contiennent toutefois des informations hiérarchiques et multicouches, et il est essentiel de les visualiser de manière séquentielle ou progressive. De plus, les représentations 3D et 2D, ainsi que les prédictions d’apprentissage automatique, doivent être connectées visuellement et interactivement dans le système.Dans cette thèse, le problème du lignage cellulaire des embryons de plantes a été le leitmotiv pour concevoir et étudier un système de visualisation qui utilise des combinaisons de représentations 3D et 2D ainsi que des visualisations pour l’apprentissage automatique. Nous avons d’abord étudié les techniques d’interaction pour la sélection 3D au sein d’un embryon de plante. Les cellules d’un embryon de plante sont jointives et constituent un ensemble d’objets 3D dense dans toutes les dimensions spatiales. Nous avons mené une étude pour évaluer trois techniques de sélection différentes, et nous avons montré que la combinaison de la technique de Sélection par Explosion et de la technique de Sélection par Liste fonctionne bien pour désigner et observer les cellules d’un embryon. Ces techniques peuvent égale- ment être étendues à d’autres données 3D denses et similaires. Deuxièmement, nous avons conçu un système de visualisations et d’interaction combiné afin que les biologistes puissent examiner les cellules de l’embryon et enregistrer l’histoire du développement dans l’arbre de lignage hiérarchique. Nous prenons en charge la construction de la hiérarchie dans deux directions, à la fois en construisant l’historique de haut en bas de l’arbre en utilisant la sélection lasso dans l’environnement 3D et de bas en haut selon le flux de travail traditionnel pour construire un arbre de lignage cellulaire hiérarchique. Nous avons également ajouté un modèle de réseau neuronal pour fournir aux biologistes des prédictions initiales sur les filiations. Nous avons réalisé une évaluation avec des biologistes ; celle- ci a montré que les représentations 3D et 2D facilitent les prises de décisions et que l’outil peut enrichir leur vision des embryons. Cependant, la performance du modèle d’apprentissage automatique n’était pas idéale. Aussi, pour faciliter le processus et améliorer les performances du modèle, dans une version plus aboutie de notre système, nous avons entraîné cinq modèles de classification différents, visualisé les prédictions et leurs incertitudes associées. Nous avons réalisé une évaluation auprès des utilisateurs ; les résultats ont indiqué que les représentations des classifieurs que nous avons conçues sont faciles à comprendre, et que le nouvel outil peut améliorer significativement les prises de décision pour la validation du lignage cellulaire.

Cultural transmission of reproductive success has been observed in many human populations as well as other animals. Cultural transmission of reproductive success consists of a positive correlation of nongenetic origin between the progeny size of parents and children. This correlation can result from various factors, such as the social influence of parents on their children, the increase of children’s survival through allocare from uncles and aunts, or the transmission of resources. Here, we study the evolution of genomic diversity over time under cultural transmission of reproductive success. Cultural transmission of reproductive success has a threefold impact on population genetics: (1) the effective population size decreases when cultural transmission of reproductive success starts, mimicking a population contraction, and increases back to its original value when cultural transmission of reproductive success stops; (2) coalescent tree topologies are distorted under cultural transmission of reproductive success, with higher imbalance and a higher number of polytomies; and (3) branch lengths are reduced nonhomogenously, with a higher impact on older branches. Under long-lasting cultural transmission of reproductive success, the effective population size stabilizes but the distortion of tree topology and the nonhomogenous branch length reduction remain, yielding U-shaped site frequency spectra under a constant population size. We show that this yields a bias in site frequency spectra-based demographic inference. Considering that cultural transmission of reproductive success was detected in numerous human and animal populations worldwide, one should be cautious because inferring population past histories from genomic data can be biased by this cultural process.

International audience; The Vador Project started in 2018 between the French Public Employment Service Pole Emploi, computer scientists from Paris Saclay and economist from CREST. The project aims to learn a new recommender system from past hirings and job seekers and offers data. We will present both the past, current and futur research and technical challenges faced by the projects and its contributions.

International audience; Algorithmic recommendations of job ads to job seekers promise to alleviate frictional unemployment, but raise fairness considerations due to biases in training data. This paper strives to discuss the issue of algorithmic fairness, with a focus on gender, in a hybrid job recommender system trained on past hires developed in partnership with the French Public Employment Service.

International audience; The Cultural Transmission of Reproductive Success (CTRS) is one of the various cultural processes that can impact human genetic evolution. In this process, individuals from large families have more children on average. Here, we develop and evaluate methods to infer this process from genomic data, using two approaches: (1) Approximate Bayesian computation, which uses summary statistics computed on inferred genealogies from genomic data and (2) deep neural networks, which are directly trained on genomic data. These methods rely on large simulated datasets incorporating varying levels of CTRS. Both competing approaches show a good ability to infer CTRS on genomic data and worth investigating under more complex evolutionary histories.

International audience; We conduct a user study to quantify and compare user performance for a value comparison task using four bar chart designs, where the bars show the mean values of data loaded progressively and updated every second (progressive bar charts).Progressive visualization divides different stages of the visualization pipeline—data loading, processing, and visualization—into iterative animated steps to limit the latency when loading large amounts of data. An animated visualization appearing quickly, unfolding, and getting more accurate with time, enables users to make early decisions. However, intermediate mean estimates are computed only on partial data and may not have time to converge to the true means, potentially misleading users and resulting in incorrect decisions.To address this issue, we propose two new designs visualizing the history of values in progressive bar charts, in addition to the use of confidence intervals. We comparatively study four progressive bar chart designs: with/without confidence intervals, and using near-history representation with/without confidence intervals, on three realistic data distributions. We evaluate user performance based on the percentage of correct answers (accuracy), response time, and user confidence. Our results show that, overall, users can make early and accurate decisions with 92% accuracy using only 18% of the data, regardless of the design. We find that our proposed bar chart design with only near-history is comparable to bar charts with only confidence intervals in performance, and the qualitative feedback we received indicates a preference for designs with history.

International audience; Physical data visualizations tap into our lifelong experience of perceiving and manipulating the physical world, either alone or with other people. However, most physical visualizations are either monolithic and static, or require human intervention to be rearranged. We drew inspiration from existing physical interactive systems and data storytelling practices with physical tokens (like those of the influential Swedish professor Hans Rosling) to develop Zooids, tiny wheeled robots that can move rapidly on any horizontal surface. We used Zooids to explore physical data visualizations that can (1) be manipulated by humans, and (2) update themselves through computerized mechanisms.

International audience; We introduce a conceptual model for scalability designed for visualization research. With this model, we systematically analyze over 120 visualization publications from 1990-2020 to characterize the different notions of scalability in these works. While many papers have addressed scalability issues, our survey identifies a lack of consistency in the use of the term in the visualization research community. We address this issue by introducing a consistent terminology meant to help visualization researchers better characterize the scalability aspects in their research. It also helps in providing multiple methods for supporting the claim that a work is "scalable". Our model is centered around an effort function with inputs and outputs. The inputs are the problem size and resources, whereas the outputs are the actual efforts, for instance, in terms of computational run time or visual clutter. We select representative examples to illustrate different approaches and facets of what scalability can mean in visualization literature. Finally, targeting the diverse crowd of visualization researchers without a scalability tradition, we provide a set of recommendations for how scalability can be presented in a clear and consistent way to improve fair comparison between visualization techniques and systems and foster reproducibility.

We anticipate a future in which people frequently have virtual content displayed in their field of view to augment reality. Situations where this virtual content interferes with users' perception of the physical world will thus be more frequent, with consequences ranging from mere annoyance to serious injuries. We argue for the need to give users agency over virtual augmentations, discussing the concept of de-augmenting augmented reality by selectively removing virtual content from the field of view. De-augmenting lets users target what actually interferes with their perception of the environment while keeping what is of interest. We contribute a framework that captures the different facets of de-augmentation. We discuss what it entails in terms of technical realization and interaction design, and end with three scenarios to illustrate what the user experience could be in a sample of domestic and professional situations.; Nous anticipons un avenir dans lequel les gens auront fréquemment du contenu virtuel affiché dans leur champ de vision pour augmenter la réalité. Les situations où ce contenu virtuel interfère avec la perception du monde physique par les utilisateurs seront donc plus fréquentes, avec des conséquences allant du simple désagrément à des accidents graves. Nous soutenons la nécessité de donner aux utilisateurs la possibilité d'agir sur les augmentations virtuelles, en discutant du concept de dé-augmentation de la réalité augmentée par la suppression sélective du contenu virtuel du champ de vision. La dé-augmentation permet aux utilisateurs de cibler ce qui interfère réellement avec leur perception de l'environnement tout en conservant ce qui est intéressant. Nous proposons un cadre qui capture les différentes facettes de la dé-augmentation. Nous discutons de ce que cela implique en termes de réalisation technique et de conception de l'interaction, et nous terminons par trois scénarios pour illustrer ce que pourrait être l'expérience de l'utilisateur dans des situations domestiques et professionnelles.

This thesis aims at identifying theoretically and concretely how visual analytics can support historians in their social network analysis process. Historical social network analysis is a method to study social relationships between groups of actors (families, institutions, companies, etc.) through a reconstruction of relationships of the past from historical documents, such as marriage acts, migration forms, birth certificates, and censuses. The use of visualization and analytical methods lets social historians explore and describe the social structure shaping those groups while explaining sociological phenomena and individual behaviors through computed network measures. However, the inspection and encoding of the sources leading to a finalized network is intricate and often results in inconsistencies, errors, distortions, and traceability problems, and current visualization tools typically have usability and interpretability issues. For these reasons, social historians are not always able to make thorough historical conclusions: many studies consist of qualitative descriptions of network drawings highlighting the presence of motifs such as cliques, components, bridges, etc. The goal of this thesis is therefore to propose visual analytics tools integrated into the global social historians' workflow, with guided and easy-to-use analysis capabilities. From collaborations with historians, I formalize the workflow of historical network analysis starting at the acquisition of sources to the final visual analysis. By highlighting recurring pitfalls, I point out that tools supporting this process should satisfy traceability, simplicity, and document reality principles to ease bask and forth between the different steps, provide tools easy to manipulate, and not distort the content of sources with modifications and simplifications. To satisfy those properties, I propose to model historical sources into bipartite multivariate dynamic social networks with roles as they provide a good tradeoff of simplicity and expressiveness while modeling explicitly the documents, hence letting users encode, correct, and analyze their data with the same abstraction and tools. I then propose two interactive visual interfaces to manipulate, explore, and analyze this data model, with a focus on usability and interpretability. The first system ComBiNet allows an interactive exploration leveraging the structure, time, localization, and attributes of the data model with the help of coordinated views and a visual query system allowing users to isolate interesting groups and individuals, and compare their position, structures, and properties. It also lets them highlight erroneous and inconsistent annotations directly in the interface. The second system, PK-Clustering, is a concrete proposition to enhance the usability and effectiveness of clustering mechanisms in social network visual analytics systems. It consists in a mixed-initiative clustering interface that let social scientists create meaningful clusters with the help of their prior knowledge, algorithmic consensus, and interactive exploration of the network. Both systems have been designed with continuous feedback from social historians, and aim to increase the traceability, simplicity, and document reality of visual analytics supported historical social network research. I conclude with discussions on the potential merging of both tools, and more globally on research directions towards better integration of visual analytics systems on the whole workflow of social historians. Systems with a focus on those properties---traceability, simplicity, and document reality---can limit the introduction of bias while lowering the requirements for the use of quantitative methods for historians and social scientists which has always been a controversial discussion among practitioners.; Cette thèse vise à identifier théoriquement et concrètement comment l'analyse visuelle peut aider les historiens dans leur processus d'analyse de réseaux sociaux. L'analyse de réseaux sociaux est une méthode utilisée en histoire sociale qui vise à étudier les relations sociales au sein de groupes d'acteurs (familles, institutions, entreprises, etc.) en reconstruisant les relations du passé à partir de documents historiques, tels que des actes de mariages, des actes de naissances, ou des recensements. L'utilisation de méthodes visuelles et analytiques leurs permet d'explorer la structure sociale formant ces groupes et de relier des mesures structurelles à des hypothèses sociologiques et des comportements individuels. Cependant, l'inspection, l'encodage et la modélisation des sources menant à un réseau finalisé donnent souvent lieu à des erreurs, des distorsions et des problèmes de traçabilité, et les systèmes de visualisation actuels présentent souvent des défauts d'utilisabilité et d'interprétabilité. En conséquence, les historiens ne sont pas toujours en mesure de faire des conclusions approfondies à partir de ces systèmes : beaucoup d'études se limitent à une description qualitative d'images de réseaux, surlignant la présence de motifs d'intérêts (cliques, îlots, ponts, etc.). Le but de cette thèse est donc de proposer des outils d'analyse visuelle adaptés aux historiens afin de leur permettre une meilleur intégration de leur processus global et des capacités d'analyse guidées. En collaboration avec des historiens, je formalise le processus d'une analyse de réseau historique, de l'acquisition des sources jusqu'à l'analyse finale, en posant comme critère que les outils utilisés dans ce processus devraient satisfaire des principes de traçabilité, de simplicité et de réalité documentaire (i.e., que les données présentées doivent être conformes aux sources) pour faciliter les va-et-vient entre les différentes étapes et la prise en main par l'utilisateur et ne pas distordre le contenu des sources. Pour satisfaire ces propriétés, je propose de modéliser les sources historiques en réseaux sociaux bipartis multivariés dynamiques avec rôles. Ce modèle intègre explicitement les documents historiques sous forme de nœuds, ce qui permet aux utilisateurs d'encoder, de corriger et d'analyser leurs données avec les mêmes outils. Je propose ensuite deux interfaces d'analyse visuelle permettant, avec une bonne utilisabilité et interprétabilité, de manipuler, d'explorer et d'analyser ce modèle de données. Le premier système ComBiNet offre une exploration visuelle de l'ensemble des dimensions du réseau à l'aide de vues coordonnées et d'un système de requêtes visuelles permettant d'isoler des individus ou des groupes et de comparer leurs structures topologiques et leurs propriétés. L'outil permet également de détecter les motifs inhabituels et ainsi de déceler les éventuelles erreurs dans les annotations. Le second système, PK-Clustering, est une proposition d'amélioration de l'utilisabilité et de l'efficacité des mécanismes de clustering dans les systèmes de visualisation de réseaux sociaux. L'interface permet de créer des regroupements pertinents à partir des connaissances a priori de l'utilisateur, du consensus algorithmique et de l'exploration du réseau dans un cadre d'initiative mixte. Les deux systèmes ont été conçus à partir des besoins et retours continus d'historiens, et visent à augmenter la traçabilité, la simplicité, et la réalité documentaire des sources dans le processus d'analyse de réseaux historiques. Je conclus sur la nécessité d'une meilleure intégration des systèmes d'analyse visuelle dans le processus de recherche des historiens. Cette intégration nécessite des outils plaçant les utilisateurs au centre du processus avec un accent sur la flexibilité et l'utilisabilité, limitant ainsi l'introduction de biais et les barrières d'utilisation des méthodes quantitatives, qui subsistent en histoire.

Inferring causal relationships from observational data is a fundamental yet highly complex problem when the number of variables is large. Recent advances have made much progress in learning causal structure models (SEMs) but still face challenges in scalability. This paper aims to efficiently discover causal DAGs from high-dimensional data. We investigate a way of recovering causal DAGs from inverse covariance estimators of the observational data. The proposed algorithm, called ICID (inverse covariance estimation and {\it independence-based} decomposition), searches for a decomposition of the inverse covariance matrix that preserves its nonzero patterns. This algorithm benefits from properties of positive definite matrices supported on {\it chordal} graphs and the preservation of nonzero patterns in their Cholesky decomposition; we find exact mirroring between the support-preserving property and the independence-preserving property of our decomposition method, which explains its effectiveness in identifying causal structures from the data distribution. We show that the proposed algorithm recovers causal DAGs with a complexity of $O(d^2)$ in the context of sparse SEMs. The advantageously low complexity is reflected by good scalability of our algorithm in thorough experiments and comparisons with state-of-the-art algorithms.

International audience

International audience; Players in Massively Multiplayer Online Role-Playing Games (MMORPGs) generate long-standing histories with their characters, but cannot express or see traces of their adventures in the game worlds. We are interested in designing game systems where players shape and contribute their own narratives as game content. Study one designed and playtested three Virtual Tabletop Role-Playing Game (VTTRPG) prototypes where we found that structured, graphical representations of players' traces support co-design and narrative analysis. We also identified four categories of traces: environment, build, memory and object. We introduce Play Traces, a novel analysis method for representing and co-designing with players and their narratives. A structured observation including Play Traces studied 17 players over 16 four-hour sessions in the third VTTRPG prototype. We found that players successfully (and enjoyably) co-designed novel narratives. We identified three themes for how traces can affect and support players in shaping new interactive narratives. We present four design implications describing how player-created narratives in MMORPGs should first Reveal & Pull Attention from other players, Invite & Push further exploration, Guide & Assist toward endings, and optionally Show & Hide traces. Finally, we discuss how treating players as co-designers offers a promising approach for developing the next generation of MMORPGs. CCS Concepts: • Human-centered computing → Human computer interaction (HCI); • Information systems → Massively multiplayer online games.

Existing systems dealing with the increasing volume of data series cannot guarantee interactive response times, even for fundamental tasks such as similarity search. Therefore, it is necessary to develop analytic approaches that support exploration and decision making by providing progressive results, before the final and exact ones have been computed. Prior works lack both efficiency and accuracy when applied to large-scale data series collections. We present and experimentally evaluate ProS, a new probabilistic learning-based method that provides quality guarantees for progressive Nearest Neighbor (NN) query answering. We develop our method for k-NN queries and demonstrate how it can be applied with the two most popular distance measures, namely, Euclidean and Dynamic Time Warping (DTW). We provide both initial and progressive estimates of the final answer that are getting better during the similarity search, as well suitable stopping criteria for the progressive queries. Moreover, we describe how this method can be used in order to develop a progressive algorithm for data series classification (based on a k-NN classifier), and we additionally propose a method designed specifically for the classification task. Experiments with several and diverse synthetic and real datasets demonstrate that our prediction methods constitute the first practical solutions to the problem, significantly outperforming competing approaches. This paper was published in the VLDB Journal (2022).

International audience; Mobile devices such as smartphones and tablets have limited display size and input capabilities that make a variety of tasks challenging. Coupling the mobile device with Augmented Reality eyewear such as smartglasses can help address some of these challenges. In the specific context of digital content annotation tasks, this combination has the potential to enhance the user experience on two fronts. First, annotations can be offloaded into the air around the mobile device, freeing precious screen real-estate. Second, as smartglasses often come equipped with a variety of sensors including a camera, users can annotate documents with pictures or videos of their environment, captured on the spot, hands-free, and from the wearer's perspective. We present AnnotAR, a prototype that we use as a research probe to assess the viability of this approach to digital content annotation. We use AnnotAR to gather users' preliminary feedback in a laboratory setting, and to showcase how it could support real-world use cases.