Ran Dubin

I am an Associate Professor in the Department of Computer and Software Engineering at Ariel University, Israel, and the founder and head of the ByteDefend Cyber Lab. My research sits at the intersection of cybersecurity, machine learning, and network intelligence.

Research Interests:

🔐 AI Model Security — detecting and disarming hidden malware, steganographic attacks, and adversarial payloads inside neural network models
🌐 Encrypted Network Traffic Analysis — fingerprinting, classification, and anomaly detection in TLS/HTTPS/QUIC traffic without decryption
🔑 API Security — few-shot and retrieval-based detection of API injection and abuse
🛡️ Content Disarm & Reconstruction (CDR) — zero-trust file sanitization for PDF, RTF, images, and AI model formats
📡 Network Anomaly & Intrusion Detection — GNN-based detection of attacks, malware C2 traffic, and cloud-service anomalies

I am passionate about translating academic research into deployable cybersecurity solutions. My group publishes at top venues including IEEE TIFS, IEEE Access, Computers & Security, IEEE ICC, and CCNC.

Google Scholar ByteDefend Lab GitHub

news

Feb 2026	📄 New preprint: SafePickle — Robust and Generic ML Detection of Malicious Pickle-based ML Models. Our latest work defends against supply-chain attacks hiding malware in serialised AI model files. 🔒
Jan 2026	🏆 Three papers accepted at IEEE CCNC 2026: ultra-fast network throughput estimation, GNN-based cloud anomaly detection, and QoE prediction for online gaming traffic.
Sep 2025	🛡️ Paper accepted at FedCSIS 2025: AI-MTD — Zero-Trust AI Model Security Based on Moving Target Defense. New paradigm for protecting deployed ML models at runtime.
Jun 2025	🌐 Four papers presented at IEEE ICC 2025: PQClass (post-quantum traffic classification), D-MAGIC (GNN-based attack detection), encrypted traffic data augmentation, and one-shot file type detection.
Jan 2025	🔑 Journal paper published in Computers & Security: A Classification-by-Retrieval Framework for Few-Shot Anomaly Detection to Detect API Injection. First retrieval-based approach to API security with minimal labelled samples.

selected publications

2026	arXiv	SafePickle: Detecting Hidden Malware in Hugging Face AI Models Ran Dubin arXiv preprint
2025	C&S	A Classification-by-Retrieval Framework for Few-Shot Anomaly Detection to Detect API Injection Udi Aharon, Ran Dubin, Amit Dvir, Chen Hajaj Computers & Security
2025	ICC	PQClass: Classification of Post-Quantum Encryption Applications in Internet Traffic Angelos K. Marnerides, Chen Hajaj, Revital Marbel, Ran Dubin, Amit Dvir IEEE International Conference on Communications (ICC)
2025	FedCSIS	AI-MTD: Zero-Trust AI Model Security Based on Moving Target Defense Daniel Gilkarov, Ran Dubin FedCSIS 2025
2023	IEEE TIFS	Content Disarm and Reconstruction of RTF Files: A Zero File Trust Methodology Ran Dubin IEEE Transactions on Information Forensics and Security
2023	IEEE Access	Disarming Attacks Inside Neural Network Models Ran Dubin IEEE Access
2023	IEEE Access	Content Disarm and Reconstruction of PDF Files Ran Dubin IEEE Access
2020	C&S	Encrypted Video Traffic Clustering Demystified Amit Dvir, Angelos K. Marnerides, Ran Dubin, Nehor Golan, Chen Hajaj Computers & Security
2017	IEEE TIFS	I Know What You Saw Last Minute — Encrypted HTTP Adaptive Video Streaming Title Classification Ran Dubin, Amit Dvir, Ofir Pele, Ofer Hadar IEEE Transactions on Information Forensics and Security

View all publications → | Google Scholar