SecureScan: AI Framework for Malware and Phishing Detection
SecureScan is an advanced detection system utilizing AI, featuring a three-tiered approach to overcome the shortcomings of conventional signature-based intrusion detection methods against advanced malware and phishing attacks. This framework employs logistic regression classification, heuristic evaluation, and external threat intelligence sourced from the VirusTotal API to analyze URLs, file hashes, and binaries. Its design processes known threats through heuristics, utilizes machine learning for uncertain samples, and confirms ambiguous cases with third-party data. In tests, SecureScan demonstrated a 93.1% accuracy rate, with balanced precision at 0.87 and recall at 0.92, showcasing excellent generalization and minimized overfitting through calibrated thresholds. A gray-zone logic (0.45-0.55) was implemented to address unclear cases, emphasizing both efficiency and effectiveness in identifying contemporary cyber threats.
Key facts
- SecureScan is an AI-driven, triple-layer detection framework.
- It integrates logistic regression, heuristic analysis, and VirusTotal API.
- Targets URLs, file hashes, and binaries.
- Achieves 93.1% accuracy on benchmark datasets.
- Precision is 0.87 and recall is 0.92.
- Uses threshold-based decision calibration and gray-zone logic (0.45-0.55).
- Addresses limitations of traditional signature-based systems.
- Framework filters known threats via heuristics and classifies uncertain samples with ML.
Entities
Institutions
- VirusTotal