Honorable Mention Long Paper Award

November 7, 2025; Christina Warren

Congratulations to Daniel Klötzl, Ozan Tastekin, David Hägele (A01), Marina Evers, and Daniel Weiskopf (A01, B01, INF, MGK, Ö)! At the IEEE Workshop on Uncertainty Visualization: Unraveling Relationships of Uncertainty, AI, and Decision-Making their paper "Uncertainty-Aware PCA for Arbitrarily Distributed Data Modeled by Gaussian Mixture Models" won the Honorable Mention Long Paper Award. The workshop was held in conjunction with IEEE VIS 2025, which took place from November 2 to 7, 2025, in Vienna.

View paper preprint on arXiv

Abstract:

Multidimensional data is often associated with uncertainties that are not well-described by normal distributions. In this work, we describe how such distributions can be projected to a low-dimensional space using uncertainty-aware principal component analysis (UAPCA). We propose to model multidimensional distributions using Gaussian mixture models (GMMs) and derive the projection from a general formulation that allows projecting arbitrary probability density functions. The low-dimensional projections of the densities exhibit more details about the distributions and represent them more faithfully compared to UAPCA mappings. Further, we support including user-defined weights between the different distributions, which allows for varying the importance of the multidimensional distributions. We evaluate our approach by comparing the distributions in low-dimensional space obtained by our method and UAPCA to those obtained by sample-based projections.

Fig.1: Overview of our uncertainty-aware PCA approach for an artificial dataset that consists of two classes of non-normally distributed samples. Each class is fitted with individual GMMs Gi and aggregated for the projection via weighted UAPCA (with given weights τ(i)). Our approach is compared to PCA with KDE on the samples and UAPCA.