通过因果引导对抗生成，提升反事实图像解释的准确性与真实性
Counterfactual Visual Explanation via Causally-Guided Adversarial Steering
- 引入因果引导的对抗方法，避免错误扰动无关因素
- 在多个数据集上优于现有方法，兼顾有效性、稀疏性、接近度和真实感
- 适合关注AI可解释性与图像生成质量的研究者
近年来，反事实视觉解释通过视觉扰动改变模型预测，提升了人工智能模型的可解释性。然而，这些方法忽略了图像生成过程中的因果关系和虚假相关性，常导致反事实图像出现意外变化，解释质量受限。为此，我们提出新框架CECAS，首先采用因果引导的对抗方法生成反事实解释，创新性地引入因果视角，避免在反事实图像中对虚假因素产生不必要的扰动。大量实验表明，该方法在多个基准数据集上超越现有最先进方法，最终在有效性、稀疏性、接近度和真实感之间实现良好平衡。
原文摘要 · Abstract (English)
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.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。