MACHINE LEARNING-BASED SECURITY VULNERABILITY DETECTION FRAMEWORK FOR CONTINUOUS SOFTWARE TESTING

Abstract

Software security vulnerabilities continue to pose significant challenges for modern software development due to increasing system complexity, frequent releases, and evolving cyber threats. Traditional vulnerability detection approaches based on manual code review, static analysis, and rule-based security testing often struggle to identify sophisticated vulnerabilities across large-scale software systems. This paper proposes a Machine Learning-Based Security Vulnerability Detection Framework for Continuous Software Testing by integrating machine learning, automated software testing, static and dynamic analysis, DevSecOps practices, and intelligent security analytics. The proposed framework continuously analyzes source code, software behavior, dependency information, and testing outcomes to identify potential security weaknesses throughout the software development lifecycle. Machine learning models classify vulnerabilities, detect anomalous coding patterns, and prioritize security risks based on severity and impact. Experimental evaluation demonstrates improvements in vulnerability detection accuracy, testing efficiency, false-positive reduction, and continuous security monitoring capability. The proposed framework provides a scalable solution for integrating intelligent vulnerability detection into modern software delivery pipelines.