Figure 1. System Architecture

Malware continues to pose a significant threat to modern computing systems due to its rapidly evolving nature and the increasing sophistication of cyber-attacks. Traditional detection approaches such as signature-based antivirus systems struggle to identify new and unknown malware variants, particularly zero-day and polymorphic threats. This paper presents a machine learning-based malware detection framework designed to classify executable files as either benign or malicious using behavioural and metadata-based features. A dataset consisting of 10,000 executable samples, including 6,000 malicious files and 4,000 benign files, was collected from public malware repositories and trusted system environments. Several machine learning algorithms, including Decision Tree, Random Forest, and Neural Network models, were implemented and evaluated. The dataset was divided into 70% training, 15% validation, and 15% testing sets to ensure reliable model evaluation. Experimental results show that the Decision Tree classifier achieved an accuracy of 88.5%, while the Random Forest model achieved 95.6% accuracy. The Neural Network model achieved the best performance with an accuracy of 96.4%, precision of 0.95, recall of 0.97, and F1-score of 0.96. These results demonstrate that machine learning techniques significantly improve malware detection performance compared to traditional signature-based approaches. The proposed system also includes a lightweight interface for file analysis, enabling efficient malware classification. Future work will focus on integrating deep learning models and real time behavioural monitoring to enhance detection of advanced malware variants.