Implementation of backdoor attacks and defenses in malware classification using machine learning models.

End-to-end PE malware detection with XGBoost and MalConv2. Adversarial robustness evaluation via GAMMA attack, SHAP interpretability, and multi-model Pareto comparison.

An end-to-end malware detection pipeline leveraging multiple machine learning models, ensemble learning, and explainable AI techniques to accurately classify malicious and benign files. Built using the EMBER 2018 dataset with XGBoost, LightGBM, CatBoost, Neural Networks, and SHAP-based interpretability.

LightGBM and Random Forest based malware detection on the EMBER 2018 dataset

OACSP-Forensics: a cross-modal selective prediction benchmark for digital forensic triage. Applies ordinal-aware class-conditional selective prediction to phishing URLs (PhishTank), malware binaries (EMBER), and memory forensics (CIC-MalMem-2022) with Daubert-aligned evaluation and coverage-risk curves for court-admissible AI triage.

Static malware detection system using Random Forest on EMBER features for offline, explainable threat analysis

2,899 real-world malware families categorized for security teams & incident response. Schema.org-ready dataset derived from EMBER 2018 with FAQ, MITRE ATT&CK, CISA advisory cross-refs, and per-family profiles. Apache-2.0 licensed.