Artificial intelligence-based detection and analysis of malware attacks using feature selection

Today's world, cybersecurity attack methods and threats have become increasingly complex, with evolving attack techniques presenting a significant challenge on a global scale. The constant evolution of malware, which is among the most critical current threats, and the development of sophisticated attack techniques highlight the inadequacy of traditional security methods. This thesis focuses on enhancing the effectiveness of malware detection and analysis processes by leveraging artificial intelligence and machine learning techniques, thereby enabling more rapid and effective responses to cyber threats. Within the scope of the study, a comprehensive dataset comprising various types of malware and attack vectors has been created, and this dataset has undergone preprocessing to ensure its suitability for analysis. During the machine learning phase, different algorithms such as decision trees, random forest, k-nearest neighbors (k-NN), and neural networks have been utilized, followed by feature selection and optimization techniques to compare model performances. The dataset was divided into two groups for testing and analysis, allowing for a thorough evaluation of the models. The best-performing model was subsequently analyzed in detail, and the obtained results were discussed comprehensively.