Concolic Testing for Deep Neural Networks

TL;DR

This paper introduces a concolic testing approach tailored for Deep Neural Networks, combining execution and symbolic analysis to improve test coverage and identify adversarial examples.

Contribution

It is the first to formalize coverage criteria for DNNs and develop a concolic testing method to enhance testing effectiveness.

Findings

Achieves high coverage of DNNs

Successfully finds adversarial examples

Demonstrates effectiveness through experiments

Abstract

Concolic testing combines program execution and symbolic analysis to explore the execution paths of a software program. This paper presents the first concolic testing approach for Deep Neural Networks (DNNs). More specifically, we formalise coverage criteria for DNNs that have been studied in the literature, and then develop a coherent method for performing concolic testing to increase test coverage. Our experimental results show the effectiveness of the concolic testing approach in both achieving high coverage and finding adversarial examples.