collaborators

28 papers

cs.LG2026

FLAT: Revealing Hidden Latent-Conditioned Backdoor Failures in Federated Learning

Tuan Nguyen, Sze Jue Yang, Khoa D. Doan +2

Horizontal federated learning (HFL) backdoor audits often summarize model behavior through clean accuracy (CA), mean attack success rate (ASR), or a single known-trigger test. Such…

cs.LG2026

Rethinking Molecular Graph Backdoors under Chemistry-aware Admission

Thinh T. H. Nguyen, Sze Jue Yang, Khoa D. Doan +2

Backdoor attacks on molecular graph neural networks (GNNs) are typically evaluated as abstract graph edits, but real molecular learning pipelines do not train on arbitrary graphs.…

cs.CV2026

StructSAM: Structure- and Spectrum-Preserving Token Merging for Segment Anything Models

Duy M. H. Nguyen, Tuan A. Tran, Duong Nguyen +17

Recent token merging techniques for Vision Transformers (ViTs) provide substantial speedups by reducing the number of tokens processed by self-attention, often without retraining.…

cs.CV2026

Look Before You Zoom: Adaptive Routing for the Resolution-Context Trade-off in Visual RAG

Oanh N. Tran, Thanh Quoc Hung Le, Oscar Chew +2

Vision-Language Models (VLMs) struggle as query-relevant objects become smaller. To address this, recent training-free approaches dynamically retrieve and zoom into local image reg…

cs.CV2026

FOCA: Future-Oriented Conditioning for Data-Efficient Vision-Language-Action Adaptation

Duc Minh Nguyen, Nghiem Tuong Diep, Binh Gia Nguyen +20

Vision-Language-Action (VLA) models enable general-purpose robotic control via large-scale multimodal pretraining, yet their effectiveness under few-shot imitation learning remains…

cs.LG2026

When Generator Replay Degrades: Projected Rehearsal Orchestration for Heterogeneous Federated Class-Incremental Learning

Thinh T. H. Nguyen, Khoa D. Doan, Binh T. Nguyen +2

Federated class-incremental learning (FCIL) becomes substantially harder when clients observe different label subsets, progress through tasks at different stages, and provide uneve…