IEEE VIS 2022 Best Paper Award for "Uncertainty-Aware Multidimensional Scaling"

October 18, 2022; Christina Warren

At IEEE VIS 2022, David Hägele, Tim Krake and Daniel Weiskopf were awarded the Best Paper Award for their paper "Uncertainty-Aware Multidimensional Scaling". The IEEE VIS is one of the world's largest and most important conferences in the field of visualization and was held this year as a hybrid event in Oklahoma City, USA. With the Best Paper Award, the conference honors outstanding papers from the pool of accepted papers.

 

Abstract

We present an extension of multidimensional scaling (MDS) to uncertain data, facilitating uncertainty visualization of multidimensional data. Our approach uses local projection operators that map high-dimensional random vectors to low-dimensional space to formulate a generalized stress. In this way, our generic model supports arbitrary distributions and various stress types. We use our uncertainty-aware multidimensional scaling (UAMDS) concept to derive a formulation for the case of normally distributed random vectors and a squared stress. The resulting minimization problem is numerically solved via gradient descent. We complement UAMDS by additional visualization techniques that address the sensitivity and trustworthiness of dimensionality reduction under uncertainty. With several examples, we demonstrate the usefulness of our approach and the importance of uncertainty-aware techniques.

 

 Fig.1: Uncertainty-Aware Multidimensional Scaling (UAMDS) for increasing levels of uncertainty. The top row shows 4-dimensional data sets consisting of 6 entries (color-coded). The left data set has no uncertainty, the middle data set models one entry as a normally distributed random vector, and the right data set entirely consists of multivariate normal distributions. The bottom row shows the respective UAMDS projection, where the isolines indicate the 25th to 75th percentile range of the projected (normal) distributions. The projected points of the left projection are shown in faded color in the other two plots to indicate the change due to uncertainty.

 

Further SFB-TRR 161 Contirbutions to IEEE VIS 2022

Both at IEE VIS and the workshops held in conjunction with IEEE VIS, serveral papers by SFB-TRR 161 researchers were accepted:

• Predicting User Preferences of Dimensionality Reduction Embedding Quality. Authors: Cristina Morariu, Adrien Bibal, Rene Cutura, Benoit Frenay, Michael Sedlmair
• Comparative Evaluation of Bipartite, Node-Link, and Matrix-Based Network Representations Authors: Moataz Abdelaal, Nathan D Schiele, Katrin Angerbauer, Kuno Kurzhals, Michael Sedlmair, Daniel Weiskopf
• RagRug: A Toolkit for Situated Analytics. Authors: Philipp Fleck, Aimee Sousa Calepso, Sebastian Hubenschmid, Michael Sedlmair, Dieter Schmalstieg
• Constrained Dynamic Mode Decomposition. Authors: Tim Krake, Daniel Klötzl, Bernhard Eberhardt, Daniel Weiskopf
• Relaxed Dot Plots: Faithful Visualization of Samples and Their Distribution. Authors: Nils Rodrigues, Christoph Schulz, Sören Döring, Daniel Baumgartner, Tim Krake, Daniel Weiskopf
• Situated Visual Analysis and Live Monitoring for Manufacturing. Authors: Michael Becher, Dominik Herr, Christoph Muller, Kuno Kurzhals, Guido Reina, Lena Wagner, Thomas Ertl, Daniel Weiskopf
• A Hybrid In Situ Approach for Cost Efficient Image Database Generation. Authors: Valentin Bruder, Matthew Larsen, Thomas Ertl, Hank Childs, Steffen Frey
• Toward Inclusiveness and Accessibility in Visualization Research: Speculations on Challenges, Solution Strategies, and Calls for Action (Position Paper). Authors: Katrin Angerbauer, Michael Sedlmair (BELIV workshop)
• Research Data Curation in Visualization. Authors: Dimitar Garkov, Christoph Müller, Matthias Braun, Daniel Weiskopf, Falk Schreiber (BELIV workshop)
• Power Overwhelming: Quantifying the Energy Cost of Visualisation. Authors: Christoph Müller, Moritz Heinemann, Daniel Weiskopf, Thomas Ertl (BELIV workshop)

Links:

IEEE VIS 2022 Conference