A novel approach for analyzing buffer overflow vulnerabilities in binary executables by using machine learning techniques

Abstract

While evaluating whether a software is secure or vulnerable with traditional methods; examination of security requirements, source code analysis and software security testing activities can be performed. In many cases, these activities cannot be performed by the end user due to not exist documentation of security related requirements, absence of source codes and need to expert security testing teams. When the software is in binary executable file format, we need expert systems, which accept just only binary executables as inputs to enable end-user side security analysis. In this study, we present a new method and its success, which is developed by using machine learning techniques to be used in the buffer overflow vulnerability analysis of binary executable formatted software applications.