AI辅助的智能系统安全性测试技术
AI-Assisted Security Testing Techniques for Intelligent Systems
陈永康 1李虎 1赵重阳 1况晓辉1
作者信息
- 1. 军事科学院系统工程研究院,北京 100101
- 折叠
摘要
以深度学习技术为核心的智能系统在经济社会各领域广泛应用,然而其安全性问题日益凸显.通过安全性测试及时发现智能系统存在的安全缺陷成为当前智能系统安全保障的重要途径.首先,从智能系统安全性测试需求切入,剖析并总结了智能系统存在的安全性问题及其测试难点;其次,围绕AI辅助的智能系统安全性测试,分别针对测试基准构建和测试用例生成两个安全性测试关键流程,进行了相关研究工作的对比和总结;最后,依据对现有研究工作的整理总结,探讨并展望了AI辅助的智能系统安全性测试技术的未来研究方向.
Abstract
Intelligent systems based on deep learning technologies have been widely deployed across various economic and social sectors.However,their security concerns have become increasingly prominent.Security testing has emerged as a crucial ap-proach to identify potential security vulnerabilities in intelligent systems.This study first analyzes and summarizes the security is-sues and the testing challenges of intelligent systems based on current security testing requirements.Subsequently,it proposes an AI-assisted security testing framework for intelligent systems,focusing on two key processes in security testing:benchmark con-struction and test case generation,and compares and summarizes relevant research in these areas.Finally,based on the review of existing research,this study also discusses and provides insights into future research directions for AI-assisted security testing of intelligent systems.
关键词
人工智能/智能系统/安全性测试/测试流程Key words
artificial intelligence/intelligent systems/security testing/testing procedure引用本文复制引用
出版年
2024
NETL
NSTL
维普
万方数据