# Rethinking Visual Counterfactual Explanations Through Region Constraint Source: [Rethinking Visual Counterfactual Explanations Through Region Constraint]() Collection index: [llms.txt]() Data updated: 2025-10-26T03:30:07.152Z Authors: Bartlomiej Sobieski, Jakub Grzywaczewski, Bartlomiej Sadlej, Matthew Tivnan, Przemyslaw Biecek Published: 2024-10-16T14:10:48.000Z ## Abstract Visual counterfactual explanations (VCEs) have recently gained immense popularity as a tool for clarifying the decision-making process of image classifiers. This trend is largely motivated by what these explanations promise to deliver -- indicate semantically meaningful factors that change the classifier's decision. However, we argue that current state-of-the-art approaches lack a crucial component -- the region constraint -- whose absence prevents from drawing explicit conclusions, and may even lead to faulty reasoning due to phenomenons like confirmation bias. To address the issue of previous methods, which modify images in a very entangled and widely dispersed manner, we propose region-constrained VCEs (RVCEs), which assume that only a predefined image region can be modified to influence the model's prediction. To effectively sample from this subclass of VCEs, we propose Region-Constrained Counterfactual Schr\\"odinger Bridges (RCSB), an adaptation of a tractable subclass of Schr\\"odinger Bridges to the problem of conditional inpainting, where the conditioning signal originates from the classifier of interest. In addition to setting a new state-of-the-art by a large margin, we extend RCSB to allow for exact counterfactual reasoning, where the predefined region contains only the factor of interest, and incorporating the user to actively interact with the RVCE by predefining the regions manually. ## Recorded benchmark results A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. - [Counterfactual Explanation on ImageNet Zebra - Sorrel](): Counterfactual Explanation; ImageNet Zebra - Sorrel; FID: 8. - [Counterfactual Explanation on ImageNet (Cheetah - Cougar)](): Counterfactual Explanation; ImageNet (Cheetah - Cougar); FID: 17.2. - [Counterfactual Explanation on ImageNet Egyptian Cat - Persian Cat](): Counterfactual Explanation; ImageNet Egyptian Cat - Persian Cat; FID: 23. - [Visual Counterfactual Explanation (Age) on CelebA-HQ](): Visual Counterfactual Explanation (Age); CelebA-HQ; FID: 4.92. - [Visual Counterfactual Explanation (Smile) on CelebA HQ](): Visual Counterfactual Explanation (Smile); CelebA-HQ; FID: 3.04. - [Visual Counterfactual Explanation (Age) on CelebA Standard](): Visual Counterfactual Explanation (Age); CelebA Standard; FID: 2.94. - [Visual Counterfactual Explanation (Smile) on CelebA Standard](): Visual Counterfactual Explanation (Smile); CelebA Standard; FID: 2.98. ## Original sources and implementation [Original paper]() [PDF]()