Posts by Collection

portfolio

Portfolio item number 1

Short description of portfolio item number 1

Portfolio item number 2

Short description of portfolio item number 2

publications

Privacy-Aware Risk-Adaptive Access Control in Health Information Systems using Topic Models

Published in Proceedings of the 23rd ACM Symposium on Access Control Models and Technologies (SACMAT), 2018

We design a privacy-preserving risk-adaptive access control framework for health information systems using topic modeling.

Recommended citation: W Zhang, H Li*, M Zhang, Z Lv. (2018). "Privacy-Aware Risk-Adaptive Access Control in Health Information Systems using Topic Models." SACMAT.

TULAM: Trajectory-User Linking via Attention Mechanism

Published in Science China Information Sciences, 67(1), 112103, 2024

We present TULAM, an attention-based deep learning model for accurately linking anonymous trajectories to users.

Recommended citation: H Li, S Cao, Y Chen, M Zhang, D Feng. (2024). "TULAM: Trajectory-User Linking via Attention Mechanism." Science China Information Sciences.

SeqMIA: Sequential-Metric Based Membership Inference Attack

Published in Proceedings of the 2024 ACM SIGSAC Conference on Computer and Communications Security (CCS), 2024

We propose SeqMIA, a novel sequential-metric based membership inference attack leveraging knowledge distillation.

Recommended citation: H Li, Z Li, S Wu, C Hu, Y Ye, M Zhang, D Feng, Y Zhang. (2024). "SeqMIA: Sequential-Metric Based Membership Inference Attack." ACM CCS.

RAG-leaks: Difficulty-Calibrated Membership Inference Attacks on Retrieval-Augmented Generation

Published in Science China Information Sciences, 68(6), 160102, 2025

We introduce RAG-leaks, a calibrated membership inference attack targeting retrieval-augmented generation systems.

Recommended citation: G Wang, J He, H Li*, M Zhang, D Feng. (2025). "RAG-leaks: Difficulty-Calibrated Membership Inference Attacks on Retrieval-Augmented Generation." Science China Information Sciences.

Enhanced Label-Only Membership Inference Attacks with Fewer Queries

Published in 34th USENIX Security Symposium (USENIX Security 25), 2025

We enhance label-only membership inference attacks by optimizing query strategies to reduce costs while maintaining high accuracy.

Recommended citation: H Li, Z Li, S Wu, Y Ye, M Zhang, D Feng, Y Zhang. (2025). "Enhanced Label-Only Membership Inference Attacks with Fewer Queries." USENIX Security.
Download Paper

Model Hijacking Attack in Federated Learning

Published in IEEE Transactions on Information Forensics and Security (TIFS), 2026

We reveal a novel model hijacking attack in federated learning systems where malicious clients can stealthily take control of the global model.

Recommended citation: Z Li, S Wu, R Chen, P Aditya, IE Akkus, M Vanga, M Zhang, H Li*, Y Zhang. (2026). "Model Hijacking Attack in Federated Learning." IEEE TIFS.

teaching

Teaching experience 1

Undergraduate course, University 1, Department, 2014

This is a description of a teaching experience. You can use markdown like any other post.

Teaching experience 2

Workshop, University 1, Department, 2015