There is Limited Correlation between Coverage and Robustness for Deep Neural Networks

TL;DR

This study empirically investigates the relationship between coverage criteria and robustness in deep neural networks, finding limited correlation and suggesting coverage improvements do not necessarily enhance robustness.

Contribution

It provides the largest systematic empirical analysis of 100 DNNs and 25 metrics, challenging assumptions about coverage's role in robustness.

Findings

Limited correlation between coverage and robustness.

Improving coverage does not necessarily improve robustness.

Provides a benchmark dataset and implementation for future research.

Abstract

Deep neural networks (DNN) are increasingly applied in safety-critical systems, e.g., for face recognition, autonomous car control and malware detection. It is also shown that DNNs are subject to attacks such as adversarial perturbation and thus must be properly tested. Many coverage criteria for DNN since have been proposed, inspired by the success of code coverage criteria for software programs. The expectation is that if a DNN is a well tested (and retrained) according to such coverage criteria, it is more likely to be robust. In this work, we conduct an empirical study to evaluate the relationship between coverage, robustness and attack/defense metrics for DNN. Our study is the largest to date and systematically done based on 100 DNN models and 25 metrics. One of our findings is that there is limited correlation between coverage and robustness, i.e., improving coverage does not help improve the robustness. Our dataset and implementation have been made available to serve as a benchmark for future studies on testing DNN.