Robustness via Uncertainty-aware Cycle Consistency

TL;DR

This paper introduces a probabilistic unpaired image translation method that models uncertainty with generalized Gaussian distributions, enhancing robustness to unseen perturbations across diverse domains.

Contribution

The paper proposes UGAC, a novel uncertainty-aware cycle consistency framework that explicitly models residuals with heavy-tailed distributions for improved robustness.

Findings

Outperforms state-of-the-art methods on various datasets.

Demonstrates increased robustness to unseen perturbations.

Effective in natural and medical imaging tasks.

Abstract

Unpaired image-to-image translation refers to learning inter-image-domain mapping without corresponding image pairs. Existing methods learn deterministic mappings without explicitly modelling the robustness to outliers or predictive uncertainty, leading to performance degradation when encountering unseen perturbations at test time. To address this, we propose a novel probabilistic method based on Uncertainty-aware Generalized Adaptive Cycle Consistency (UGAC), which models the per-pixel residual by generalized Gaussian distribution, capable of modelling heavy-tailed distributions. We compare our model with a wide variety of state-of-the-art methods on various challenging tasks including unpaired image translation of natural images, using standard datasets, spanning autonomous driving, maps, facades, and also in medical imaging domain consisting of MRI. Experimental results demonstrate that our method exhibits stronger robustness towards unseen perturbations in test data. Code is released here: https://github.com/ExplainableML/UncertaintyAwareCycleConsistency.