Towards Understanding Deep Learning Model in Image Recognition via Coverage Test

TL;DR

This paper empirically analyzes the relationships between different neural network coverage metrics, model depth, and dataset size across various DNN architectures to enhance security testing of deep learning models in image recognition.

Contribution

It provides the first empirical study on how coverage metrics relate to model depth, configuration, and dataset size, offering insights for improving DNN security testing.

Findings

Coverage metrics vary with model depth and configuration.

Modified decision/condition coverage relates to dataset size.

Empirical patterns identified across different DNN architectures.

Abstract

Deep neural networks (DNNs) play a crucial role in the field of artificial intelligence, and their security-related testing has been a prominent research focus. By inputting test cases, the behavior of models is examined for anomalies, and coverage metrics are utilized to determine the extent of neurons covered by these test cases. With the widespread application and advancement of DNNs, different types of neural behaviors have garnered attention, leading to the emergence of various coverage metrics for neural networks. However, there is currently a lack of empirical research on these coverage metrics, specifically in analyzing the relationships and patterns between model depth, configuration information, and neural network coverage. This paper aims to investigate the relationships and patterns of four coverage metrics: primary functionality, boundary, hierarchy, and structural coverage. A series of empirical experiments were conducted, selecting LeNet, VGG, and ResNet as different DNN architectures, along with 10 models of varying depths ranging from 5 to 54 layers, to compare and study the relationships between different depths, configuration information, and various neural network coverage metrics. Additionally, an investigation was carried out on the relationships between modified decision/condition coverage and dataset size. Finally, three potential future directions are proposed to further contribute to the security testing of DNN Models.