扩散积分梯度：用于灵活特征归因的可控路径生成
Diffusion Integrated Gradients: Controllable Path Generation for Flexible Feature Attribution
AI总结：
提出扩散积分梯度（DiffIG），通过扩散模型生成路径并嵌入用户引导，实现灵活可控的特征归因，在定量和定性上优于现有方法。
摘要：
基于路径的归因方法，如积分梯度（IG），因其强大的公理特性和通过沿基线到输入的路径积分梯度来归因模型预测到输入特征的有效性而被广泛采用。然而，归因路径的选择很大程度上影响解释的质量，现有方法依赖于固定或手工设计的路径，通常会产生噪声或失真的归因。为解决这一局限，我们提出扩散积分梯度（DiffIG），一种将路径生成重新表述为条件生成建模问题的新方法。DiffIG首先训练一个扩散模型来学习由断棍过程生成的路径分布，然后采用引导采样在采样过程中嵌入用户引导。我们证明DiffIG在定量上匹配或优于现有的基于路径的方法，实现感知上对齐的解释。这项工作为灵活、推理时可控的可解释人工智能（XAI）方法引入了新的生成视角。
英文摘要：
Path-based attribution methods such as Integrated Gradients (IG) are widely adopted for their strong axiomatic properties and effectiveness in attributing model predictions to input features by integrating gradients along a path from a baseline to the input. However, the choice of the attribution path largely affects the quality of explanations, and existing approaches rely on fixed or hand-crafted paths that often produce noisy or distorted attributions. To address this limitation, we propose Diffusion Integrated Gradients (DiffIG), a novel method that reformulates path generation as a conditional generative modeling problem. DiffIG first trains a diffusion model to learn a distribution over paths generated from a Stick-Breaking Process, then employs guided sampling to embed user guidance during the sampling procedure. We demonstrate that DiffIG quantitatively matches or outperforms existing path-based methods, achieving perceptually aligned explanations. This work introduces a new generative perspective for flexible, inference-time controllable Explainable Artificial Intelligence (XAI) methods.
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。
