Malware Detection Using Machine Learning

Abstract

Malware detection is a critical cybersecurity task, and this research explores the application of machine learning techniques to enhance detection accuracy. Leveraging Logistic Regression, Decision Tree, and Random Forest Classifier algorithms, our approach effectively classifies files as benign or malicious based on extracted features. Feature selection is performed to identify the most informative attributes. The models are evaluated on performance metrics, including accuracy and ROC curves, demonstrating their effectiveness. By utilizing ensemble methods and interpretability of Decision Trees, we aim to provide robust, explainable, and high-accuracy malware detection solutions. In a comparative analysis, we assess the strengths and weaknesses of each algorithm, enabling practitioners to make informed choices. Furthermore, we address the challenge of handling imbalanced datasets, which is common in real-world scenarios, ensuring that our approach maintains a high detection rate for both benign and malicious samples.