A COMPARATIVE EVALUATION OF DIFFERENT MACHINE LEARNING TECHNIQUES IN MALWARE DETECTION & ANALYSIS

Abstract

The persistent evolution of malware and its growing sophistication poses a significant challenge to cybersecurity. This study undertakes a comparative analysis of several machine learning approaches—namely, clustering-based anomaly detection, supervised learning using Random Forest classifiers, and time-series analysis via LSTM networks—for the purpose of malware detection. Using a dataset of over 40,000 records with 25 features, the study evaluates feature extraction, scaling, and performance metrics including precision, recall, and F1-score. Our findings highlight the superior performance of security-related features in classification tasks and the necessity of fine-tuned LSTM models for time-dependent intrusion detection. The comparative insights aim to aid cybersecurity professionals in selecting optimal machine learning strategies for robust malware detection.