Institutionen för systemteknik (ISY)

Hinge-Wasserstein: Estimating Multimodal Aleatoric Uncertainty in Regression Tasks

Ziliang Xiong, Arvi Jonnarth, Abdelrahman Eldesokey, Joakim Johnander, Bastian Wandt, Per-Erik Forssén
SAIAD2024, Seattle, USA
Safe Artificial Intelligence for All Domains (SAIAD), CVPR-W
June 2024

Abstract

Computer vision systems that are deployed in safety-critical applications need to quantify their output uncertainty. We study regression from images to parameter values and here it is common to detect uncertainty by predicting probability distributions. In this context, we investigate the regression-by-classification paradigm which can represent multimodal distributions, without a prior assumption on the number of modes. Through experiments on a specifically designed synthetic dataset, we demonstrate that traditional loss functions lead to poor probability distribution estimates and severe overconfidence, in the absence of full ground truth distributions. In order to alleviate these issues, we propose hinge-Wasserstein -- a simple improvement of the Wasserstein loss that reduces the penalty for weak secondary modes during training. This enables prediction of complex distributions with multiple modes, and allows training on datasets where full ground truth distributions are not available. In extensive experiments, we show that the proposed loss leads to substantially better uncertainty estimation on two challenging computer vision tasks: horizon line detection and stereo disparity estimation.

Full Paper

Manuscript on ArXiv will be published by IEEE and CVF after the conference.

---

Bibtex entry

@InProceedings{xiong24,
  author = 	 {Ziliang Xiong and Arvi Jonnarth and Abdelrahman Eldesokey and Joakim Johnander and Bastian Wandt and Per-Erik Forss\'en},
  title = 	 {Hinge-Wasserstein: Estimating Multimodal Aleatoric Uncertainty in Regression Tasks},
  booktitle = {Safe Artificial Intelligence for All Domains ({SAIAD}), {CVPR-W}},
  month = {June},
  address = {Seattle, USA},
  year = {2024},
  note = {Accepted.},
  url = {https://sites.google.com/view/saiad-2024/}
}