Cognitive Cyber Security Analysis Using Machine Learning Approaches

The field of cyber security has become increasingly important as organizations strive to protect their digital assets from a growing number of threats. Identifying and mitigating these threats promptly is crucial, and one of the primary methods of doing so is through the assessment of vulnerability severity. This project focuses on using machine learning techniques to analyze cyber security vulnerabilities and predict their severity. By examining key factors such as CVSS scores (CVSS-V4, CVSS-V3, CVSS-V2), CWE identifiers, and textual descriptions of vulnerabilities, the project aims to develop a predictive model that can classify vulnerabilities into one of four severity categories: LOW, MEDIUM, HIGH, and CRITICAL. The machine learning algorithms employed in this project include CatBoost, Naive Bayes, and Deep Neural Networks (DNN). These models are trained using historical vulnerability data, evaluated using performance metrics like accuracy, precision, recall, and F1-score, and ultimately used for predicting the severity of future vulnerabilities. The project also involves the development of a web-based application where users can input vulnerability details and receive a severity classification. This tool will be useful for cyber security professionals in assessing potential threats and prioritizing remediation efforts. Through this project, the goal is to enhance decision-making in cyber security, ensuring that critical vulnerabilities are promptly addressed.

Keywords: Cyber security, machine learning, vulnerability severity, CVSS scores, CWE identifiers, CatBoost, Naive Bayes, Deep Neural Networks, classification, predictive model.