Published Paper Details:

The rapid increase in sophisticated malware poses a significant threat to modern digital systems. Traditional signature-based detection methods are ineffective against zero-day and polymorphic attacks. This paper presents "Shakti Defender AI," an intelligent malware detection system that integrates both static and dynamic analysis using machine learning techniques. The proposed system extracts features from Portable Executable (PE) files, opcode sequences, and real-time behavioral patterns such as API calls, registry activities, and network interactions. These features are processed using an ensemble of machine learning models including XGBoost, LightGBM, Random Forest, and an LSTM-based model for sequence analysis. Experimental results demonstrate high performance with an accuracy of 99.16%, along with strong precision and recall. The system is implemented as a user-friendly desktop application supporting multiple file types with real-time scanning and threat intelligence integration. The results indicate that the hybrid approach significantly improves malware detection capability compared to traditional and single-method approaches, making it suitable for real-world cybersecurity applications.