通过扩散生成反事实图像，高效发现可解释的语义概念方向。
Discovering Concept Directions from Diffusion-based Counterfactuals via Latent Clustering
- 基于真实与反事实图像的潜在差异向量聚类，提取全局概念方向。
- 相比基线方法，存储减少4.6%，速度提升5.3%，无需GPU加速。
- 适用于医疗图像等高风险领域，能揭示临床特征与数据偏见。
概念解释已成为可解释人工智能中的有效方法，通过将模型决策与人类可理解的概念对齐提供可解释性洞察。然而，现有方法依赖计算密集型过程，难以高效捕捉复杂语义概念。本文提出基于潜在聚类的概念方向（CDLC），通过聚类真实图像与扩散生成的反事实图像之间的潜在差异向量，提取全局、类别特定的概念方向。相较于基线方法，CDLC在存储上减少约4.6%，在概念发现速度上提升约5.3%，且聚类过程无需使用GPU，从而实现跨潜在维度的多维语义概念高效提取。该方法在真实皮肤病变数据集上验证，所提取的概念方向与临床上公认的皮肤镜特征一致，并在某些情况下揭示了数据集特异性偏差或未知生物标志物。结果表明，CDLC具有可解释性、可扩展性，适用于高风险领域和多种数据模态。
原文摘要 · Abstract (English)
Concept-based explanations have emerged as an effective approach within Explainable Artificial Intelligence, enabling interpretable insights by aligning model decisions with human-understandable concepts. However, existing methods rely on computationally intensive procedures and struggle to efficiently capture complex, semantic concepts. This work introduces the Concept Directions via Latent Clustering (CDLC), which extracts global, class-specific concept directions by clustering latent difference vectors derived from factual and diffusion-generated counterfactual image pairs. CDLC reduces storage requirements by ~4.6% and accelerates concept discovery by ~5.3% compared to the baseline method, while requiring no GPU for clustering, thereby enabling efficient extraction of multidimensional semantic concepts across latent dimensions. This approach is validated on a real-world skin lesion dataset, demonstrating that the extracted concept directions align with clinically recognized dermoscopic features and, in some cases, reveal dataset-specific biases or unknown biomarkers. These results highlight that CDLC is interpretable, scalable, and applicable across high-stakes domains and diverse data modalities.
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。