Learning Propagation Rules for Attribution Map Generation

TL;DR

This paper introduces a learnable propagation rule module for attribution map generation, enabling adaptive, more accurate, and visually plausible explanations across various models and datasets.

Contribution

It proposes a novel learnable plugin for automatic propagation rule learning, improving attribution map quality over handcrafted methods.

Findings

Achieves state-of-the-art attribution map results

Produces cleaner and more visually plausible explanations

Demonstrates effectiveness across multiple datasets and architectures

Abstract

Prior gradient-based attribution-map methods rely on handcrafted propagation rules for the non-linear/activation layers during the backward pass, so as to produce gradients of the input and then the attribution map. Despite the promising results achieved, such methods are sensitive to the non-informative high-frequency components and lack adaptability for various models and samples. In this paper, we propose a dedicated method to generate attribution maps that allow us to learn the propagation rules automatically, overcoming the flaws of the handcrafted ones. Specifically, we introduce a learnable plugin module, which enables adaptive propagation rules for each pixel, to the non-linear layers during the backward pass for mask generating. The masked input image is then fed into the model again to obtain new output that can be used as a guidance when combined with the original one. The introduced learnable module can be trained under any auto-grad framework with higher-order differential support. As demonstrated on five datasets and six network architectures, the proposed method yields state-of-the-art results and gives cleaner and more visually plausible attribution maps.