Security and Artificial Intelligence Lab
Trustworthy, distributed, and efficient AI research
College of Engineering and Computer Science, VinUniversity
Our research group is situated within VinUniversity's College of Engineering and Computer Science. We specialize in trustworthy, distributed, and efficient AI, with a core focus on developing machine learning systems that remain robust, private, scalable, and practical in real-world settings. Our research encompasses federated learning, privacy-preserving machine learning, backdoor attacks and defenses, communication-efficient learning, edge AI, continual learning, and resource-aware AI to enhance security, privacy, reliability, efficiency, and fairness.
News
- 2026 June 2026: WiFi-Mamba accepted to ICML 2026 - Paper accepted
Interleaved Selective State Space Models for Efficient WiFi-Based 3D Multi-Person Pose Estimation has been accepted to ICML 2026. Congratulations to Quang-Anh N. D. and Prof. Kok-Seng Wong. - 2026 June 22, 2026: Rethinking Molecular Graph Backdoors released on arXiv - arXiv preprint
Rethinking Molecular Graph Backdoors under Chemistry-aware Admission is now available on arXiv. The work studies chemistry-aware admission checks and admission-aware molecular graph backdoor attacks. - 2026 June 2026: H-SFP provisionally accepted to ECCV 2026 - Paper accepted
H-SFP: Hierarchical Federated Learning with Decoupled Split-Model Prototyping has been provisionally accepted to ECCV 2026. Congratulations to Dung Tran, Nguyen Binh Ha, Duong Nguyen, Dinh Nguyen, and Prof. Kok-Seng Wong. - 2026 June 14, 2026: Projected Rehearsal Orchestration released on arXiv - arXiv preprint
When Generator Replay Degrades: Projected Rehearsal Orchestration for Heterogeneous Federated Class-Incremental Learning is now available on arXiv. - 2026 May 2026: ICML 2026 reviewing recognition - Professional service
Prof. Kok-Seng Wong has been recognized as an ICML 2026 Gold Reviewer, and Thinh Nguyen has been recognized as an ICML 2026 Silver Reviewer. - 2026 April 2026: NeurIPS 2026 Area Chair service - Professional service
Prof. Kok-Seng Wong will serve as an Area Chair for NeurIPS 2026. - 2026 April 2026: TrustFed: Trustworthy Federated Large Language Models - Project launch
TrustFed is funded by the Accelerating Research Excellence Program, VinUniversity, from April 2026 to April 2028. Principal Investigator: Prof. Kok-Seng Wong. - 2026 March 2026: BackFed accepted at the ICLR 2026 Trustworthy AI Workshop - Paper accepted
BackFed has been accepted at the ICLR 2026 Workshop on Principled Design for Trustworthy AI. - 2026 February 2026: HFedATM accepted at CVPR 2026 - Paper accepted
HFedATM has been accepted at the main technical track of CVPR 2026. - 2026 February 2026: Two papers accepted to CVPR 2026 Findings - Paper accepted
Onboarding Without Forgetting and Memory-efficient Continual Learning with Prototypical Exemplar Condensation have been accepted at CVPR 2026 Findings. - 2025 November 2025: Clean-Label Physical Backdoor Attacks accepted at AAAI 2026 - Paper accepted
Clean-Label Physical Backdoor Attacks with Data Distillation has been accepted at the main technical track of AAAI 2026. - 2025 August 2025: SC-GIR accepted at IEEE Transactions on Mobile Computing - Paper accepted
SC-GIR: Goal-oriented Semantic Communication via Invariant Representation Learning for Image Transmission has been accepted at IEEE TMC.