Diffusion Integrated Gradients: Controllable Path Generation for Flexible Feature Attribution

Unknown Date

AI Summary

Existing path integral gradient methods rely on fixed or handcrafted attribution paths, often resulting in high noise and severe distortion. This work introduces diffusion generative models into this framework for the first time, formulating path generation as a conditional generation task. It models the path prior using a Stick-Breaking process and incorporates guided sampling to enable user-controllable path generation. The proposed approach offers flexible and controllable inference, significantly improving attribution quality. Quantitative evaluations demonstrate that it matches or surpasses current state-of-the-art methods, while the generated attributions exhibit stronger alignment with human perception.

Abstract

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.

Problem

• feature attribution
• attribution path
• Integrated Gradients
• Explainable AI
• path generation

Innovation

• Diffusion Integrated Gradients
• path generation
• feature attribution
• guided sampling
• Explainable AI