使用机器学习结合源代码的信号与自然语言处理技术，通过MARFCAT对漏洞和弱点进行指纹识别、检测与分类
密码学与安全
2011-11-08 v6 编程语言
摘要
我们提出了一种机器学习方法，用于静态代码分析和弱点指纹识别，这些弱点涉及安全、软件工程及其他方面。该方法基于开源MARF框架及其上的MARFCAT应用程序，用于NIST的SATE2010静态分析工具展示研讨会（网址：http://samate.nist.gov/SATE2010Workshop.html）。
引用
@article{arxiv.1010.2511,
title = {The use of machine learning with signal- and NLP processing of source code to fingerprint, detect, and classify vulnerabilities and weaknesses with MARFCAT},
author = {Serguei A. Mokhov},
journal= {arXiv preprint arXiv:1010.2511},
year = {2011}
}
备注
33 pages, 11 tables; some results presented at SATE2010; NIST, October 2011; shorter version of v5 appears in the NIST technical report at http://samate.nist.gov/docs/NIST_Special_Publication_500-283.pdf#page=49 where its presentation is found at http://samate.nist.gov/docs/SATE2010/SATE10_13_Marfcat_Mokhov.pdf and the MARFCAT OSS release at http://sourceforge.net/projects/marf/files/Applications/MARFCAT/