Software Vulnerability Classification based on Machine Learning Algorithm

Abstract

Vulnerabilities in security are the main issues in computer security. Throughout recent years, several strategies have been used to minimize the risk of software vulnerabilities due to their high severity impacts. Machine-learning and data-mining techniques are among other solutions to investigate such issues in different environments. In this research, we investigate a comprehensive investigation and analysis of the several approaches which work for vulnerability assessment using machine learning as well as data mining techniques in the field of software vulnerability analysis and discovery. The work proposed software bug classification and vulnerability identification form completed software code using machine learning techniques. Various pre-processing and natural Language Processing (NLP) techniques have been used to extract the features from the heterogeneous dataset and generate normalized feature vectors. Those vector passes to the training module and generates Background Knowledge (BK) respectively. Three different machine learning algorithms like SVM, ANN, Random Forest have used to detect the bugs and evaluated proposed system effectiveness with some existing researches. Finally, we conclude system provides drastic supervision and better detection accuracy which is most effective and better than other machine learning algorithms.