EuroVisShort2019

Permanent URI for this collection

https://diglib.eg.org/handle/10.2312/2632781

Browse

Browsing EuroVisShort2019 by Subject "Computing methodologies"

Now showing 1 - 3 of 3
  • Thumbnail Image
    Item
    Online Learning of Visualization Preferences through Dueling Bandits for Enhancing Visualization Recommendations
    (The Eurographics Association, 2019) Kassel, Jan-Frederik; Rohs, Michael; Johansson, Jimmy and Sadlo, Filip and Marai, G. Elisabeta
    A visualization recommender supports the user through automatic visualization generation. While previous contributions primarily concentrated on integrating visualization design knowledge either explicitly or implicitly, they mostly do not consider the user's individual preferences. In order to close this gap we explore online learning of visualization preferences through dueling bandits. Additionally, we consider this challenge from a usability perspective. Through a user study (N = 15), we empirically evaluate not only the bandit's performance in terms of both effectively learning preferences and properly predicting visualizations (satisfaction regarding the last prediction: μ = 85%), but also the participants' effort with respect to the learning procedure (e.g., NASA-TLX = 24:26). While our findings affirm the applicability of dueling bandits, they further provide insights on both the needed training time in order to achieve a usability-aligned procedure and the generalizability of the learned preferences. Finally, we point out a potential integration into a recommender system.
  • Thumbnail Image
    Item
    ReLVis: Visual Analytics for Situational Awareness During Reinforcement Learning Experimentation
    (The Eurographics Association, 2019) Saldanha, Emily; Praggastis, Brenda; Billow, Todd; Arendt, Dustin L.; Johansson, Jimmy and Sadlo, Filip and Marai, G. Elisabeta
    Reinforcement learning (RL) is a branch of machine learning where an agent learns to maximize reward through trial and error. RL is challenging and data/compute intensive leading practitioners to become overwhelmed and make poor modeling decisions. Our contribution is a Visual Analytics tool designed to help data scientists maintain situation awareness during RL experimentation. Our tool allows users to understand which hyper-parameter values lead to better or worse outcomes, what behaviors are associated with high and low reward, and how behaviors evolve throughout training. We evaluated our tool through three uses cases using state of the art deep RL models demonstrating how our tool leads to RL situation awareness.
  • Thumbnail Image
    Item
    Voronoi-Based Foveated Volume Rendering
    (The Eurographics Association, 2019) Bruder, Valentin; Schulz, Christoph; Bauer, Ruben; Frey, Steffen; Weiskopf, Daniel; Ertl, Thomas; Johansson, Jimmy and Sadlo, Filip and Marai, G. Elisabeta
    Foveal vision is located in the center of the field of view with a rich impression of detail and color, whereas peripheral vision occurs on the side with more fuzzy and colorless perception. This visual acuity fall-off can be used to achieve higher frame rates by adapting rendering quality to the human visual system. Volume raycasting has unique characteristics, preventing a direct transfer of many traditional foveated rendering techniques. We present an approach that utilizes the visual acuity fall-off to accelerate volume rendering based on Linde-Buzo-Gray sampling and natural neighbor interpolation. First, we measure gaze using a stationary 1200 Hz eye-tracking system. Then, we adapt our sampling and reconstruction strategy to that gaze. Finally, we apply a temporal smoothing filter to attenuate undersampling artifacts since peripheral vision is particularly sensitive to contrast changes and movement. Our approach substantially improves rendering performance with barely perceptible changes in visual quality. We demonstrate the usefulness of our approach through performance measurements on various data sets.