Attribution for Enhanced Explanation with Transferable Adversarial eXploration

TL;DR

AttEXplore++ significantly improves deep neural network interpretability by integrating transferable adversarial attack methods, leading to more accurate, robust, and stable model explanations across various architectures and datasets.

Contribution

This paper introduces AttEXplore++, an enhanced attribution framework that incorporates transferable adversarial attacks like MIG and GRA, improving explanation accuracy and robustness.

Findings

Achieves 7.57% average improvement over AttEXplore

Improves interpretability scores by 32.62% over other methods

Provides more stable explanations across different models

Abstract

The interpretability of deep neural networks is crucial for understanding model decisions in various applications, including computer vision. AttEXplore++, an advanced framework built upon AttEXplore, enhances attribution by incorporating transferable adversarial attack methods such as MIG and GRA, significantly improving the accuracy and robustness of model explanations. We conduct extensive experiments on five models, including CNNs (Inception-v3, ResNet-50, VGG16) and vision transformers (MaxViT-T, ViT-B/16), using the ImageNet dataset. Our method achieves an average performance improvement of 7.57\% over AttEXplore and 32.62\% compared to other state-of-the-art interpretability algorithms. Using insertion and deletion scores as evaluation metrics, we show that adversarial transferability plays a vital role in enhancing attribution results. Furthermore, we explore the impact of randomness, perturbation rate, noise amplitude, and diversity probability on attribution performance, demonstrating that AttEXplore++ provides more stable and reliable explanations across various models. We release our code at: https://anonymous.4open.science/r/ATTEXPLOREP-8435/