The objective of this project is to develop an AI-driven malware detection system capable of identifying zero-day threats. By using anomaly classification techniques, the system detects unknown malware based on deviations from normal system behavior.
Study zero-day malware characteristics and detection challenges.
Research anomaly detection algorithms such as Isolation Forest or Autoencoders.
Collect datasets representing normal system behavior.
Design feature extraction mechanisms capturing system metrics such as CPU usage, file activity, and network traffic.
Train anomaly detection models on baseline normal behavior.
Evaluate model performance against simulated unknown malware.
Implement threshold tuning to reduce false positives.
Develop dashboards showing anomaly scores and system health.
Test robustness under different attack scenarios.
Document detection effectiveness and model limitations.
Suggest improvements for production-level deployment.
