Towards Frequency-Based Explanation for Robust CNN

TL;DR

This paper explores the role of frequency components in CNN explanations, revealing that reliance on high-frequency features leads to vulnerability, and that emphasizing low-frequency features enhances robustness.

Contribution

It introduces a frequency-based perspective to CNN explanations and demonstrates how frequency reliance affects model robustness and vulnerability to adversarial attacks.

Findings

Models relying on high-frequency features are more vulnerable to adversarial attacks.

Strengthening low-frequency feature associations improves model robustness.

Adversarial training increases low-frequency feature reliance.

Abstract

Current explanation techniques towards a transparent Convolutional Neural Network (CNN) mainly focuses on building connections between the human-understandable input features with models' prediction, overlooking an alternative representation of the input, the frequency components decomposition. In this work, we present an analysis of the connection between the distribution of frequency components in the input dataset and the reasoning process the model learns from the data. We further provide quantification analysis about the contribution of different frequency components toward the model's prediction. We show that the vulnerability of the model against tiny distortions is a result of the model is relying on the high-frequency features, the target features of the adversarial (black and white-box) attackers, to make the prediction. We further show that if the model develops stronger association between the low-frequency component with true labels, the model is more robust, which is the explanation of why adversarially trained models are more robust against tiny distortions.