通过因果引导对抗性引导实现反事实视觉解释
Counterfactual Visual Explanation via Causally-Guided Adversarial Steering
AI总结：
针对反事实视觉解释忽视因果关系而易产生非预期图像改变的问题，提出 CECAS 框架，用因果引导的对抗性方法避免扰动虚假因素，并在多个基准上取得更优且均衡的解释效果。
摘要：
近期关于反事实视觉解释的研究通过提供视觉扰动来翻转预测，从而提升人工智能模型的可解释性。然而，这些方法忽视了图像生成过程背后的因果关系和虚假关联，往往导致反事实图像出现非预期的改变，并使解释质量受限。为应对这一挑战，我们提出了一个新颖框架 CECAS，该框架首先利用因果引导的对抗性方法生成反事实解释。它创新性地融入因果视角，以避免反事实中对虚假因素产生不必要的扰动。大量实验表明，我们的方法在多个基准数据集上优于现有最先进方法，并最终在有效性、稀疏性、邻近性和真实性等多个方面实现了均衡权衡。
英文摘要：
Recent work on counterfactual visual explanations has contributed to making artificial intelligence models more explainable by providing visual perturbation to flip the prediction. However, these approaches neglect the causal relationships and the spurious correlations behind the image generation process, which often leads to unintended alterations in the counterfactual images and renders the explanations with limited quality. To address this challenge, we introduce a novel framework CECAS, which first leverages a causally-guided adversarial method to generate counterfactual explanations. It innovatively integrates a causal perspective to avoid unwanted perturbations on spurious factors in the counterfactuals. Extensive experiments demonstrate that our method outperforms existing state-of-the-art approaches across multiple benchmark datasets and ultimately achieves a balanced trade-off among various aspects of validity, sparsity, proximity, and realism.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
