Secure AI Lab is an academic research laboratory focused on the intersection of artificial intelligence and cybersecurity, specifically in the areas of privacy-preserving machine learning and secure AI systems. The lab conducts fundamental and applied research on topics including: - Privacy-preserving deep learning using techniques such as Fully Homomorphic Encryption (FHE), Federated Learning (FL), Differential Privacy, and Secure Multi-Party Computation (SMC) - Secure AI inference on encrypted data without exposing plaintext - Adversarial robustness and resilience of AI models - Privacy-preserving digital forensics applications Notable research projects and publications include: - HomomorphicEncryption-based Federated Learning (HEFL): Integrating CKKS and BFV encryption schemes into FL aggregation pipelines to protect gradient exchanges - SecPATE: An enhancement of the Private Aggregation of Teacher Ensembles (PATE) framework using Secure Multi-Party Computation - Pri-WeDec: A framework for weapon detection in digital forensics using FHE-encrypted inference - Pri-Inv: A privacy-preserving face presence verification tool for criminal investigations - Multi-modal explicit content detection in digital forensics combining adversarial-resilient ensemble learning with homomorphic encryption The lab publishes research in international conferences and journals including IEEE and Springer venues. It also provides open-source code, datasets, teaching materials, and scholarship resources. The lab is contactable via a Vietnamese phone number (+84), indicating it is based in Vietnam.