DeepLogic:Priority Testing of Deep Learning Through Interpretable Logic Units

扫码查看

原文链接

万方数据

外文摘要：With the increasing deployment of deep learning-based systems in various scenes,it is becoming important to conduct sufficient testing and evaluation of deep learning models to improve their interpretability and robustness.Recent studies have proposed different criteria and strategies for deep neural network(DNN)testing.However,they rarely conduct effective testing on the robustness of DNN models and lack interpretability.This paper proposes a new priority testing criterion,called DeepLogic,to analyze the robustness of the DNN models from the perspective of model interpretability.We first define the neural units in DNN with the highest average activation probability as"interpretable logic units".We analyze the changes in these units to evaluate the model's robustness by conducting adversarial attacks.After that,the interpretable logic units of the inputs are taken as context attri-butes,and the probability distribution of the softmax layer in the model is taken as internal attributes to establish a comprehensive test prioritization framework.The weight fusion of context and internal factors is carried out,and the test cases are sorted according to this priority.The experimental results on four popular DNN models using eight testing metrics show that our DeepLogic significantly outperforms existing state-of-the-art methods.

外文关键词：

Deep learning testingInterpretable logic unitsAdversarial testModel interpretabilityDefect de-tection

作者：

Chenhao LIN、Xingliang ZHANG、Chao SHEN

展开 >

作者单位：

Faculty of Electronic and Infomation Engineering,Xi'an Jiaotong University,Xi'an 710049,China

基金：

National Key Research and Development Program of ChinaNational Natural Science Foundation of ChinaNational Natural Science Foundation of ChinaNational Natural Science Foundation of ChinaNational Natural Science Foundation of ChinaNational Natural Science Foundation of ChinaNational Natural Science Foundation of ChinaShaanxi Province Key Industry Innovation Program

项目编号：

2020AAA0107702620061816216116033762132011U21B2018U20A20177622062172021ZD LGY01-02

出版年：

2024

DOI：

10.23919/cje.2022.00.451

电子学报(英文)

CSTPCDEI

ISSN：1022-4653

年,卷(期)：2024.33(4)