Posts by Collection

Portfolio item number 1

Short description of portfolio item number 1

Portfolio item number 2

Short description of portfolio item number 2

Graph Neural Network Explanations are Fragile.

Published in ICML 2024, 2024

This paper studies the vulnerability of GNN explanations.

Recommended citation: Jiate Li, Meng Pang, Yun Dong, Jinyuan Jia, Binghui Wang. (2024). "Graph Neural Network Explanations are Fragile." ICML 2024.
Download Paper

Practicable Black-box Evasion Attacks on Link Prediction in Dynamic Graphs–A Graph Sequential Embedding Method

Published in AAAI 2025, 2024

This paper proposed a practical and effective black-box attack against LPDG.

Recommended citation: Jiate Li, Meng Pang, Binghui Wang. (2025). "Practicable Black-box Evasion Attacks on Link Prediction in Dynamic Graphs--A Graph Sequential Embedding Method." AAAI 2025.
Download Paper

AGNNCert: Defending Graph Neural Networks against Arbitrary Perturbations with Deterministic Certification

Published in Usenix Security 2025, 2025

This paper proposed an effective certifiable robust GNN method against arbitary perturbations

Recommended citation: Jiate Li, Binghui Wang. (2025). "AGNNCert: Defending Graph Neural Networks against Arbitrary Perturbations with Deterministic Certification." Usenix Security 2025.
Download Paper

Provably Robust Explainable Graph Neural Networks against Graph Perturbation Attacks

Published in ICLR 2025, 2025

This paper proposed a provably robust framework for GNN explainers.

Recommended citation: Jiate Li, Meng Pang, Yun Dong, Jinyuan Jia, Binghui Wang. (2025). "Provably Robust Explainable Graph Neural Networks against Graph Perturbation Attacks." ICLR 2025.
Download Paper

Deterministic Certification of Graph Neural Networks against Graph Poisoning Attacks with Arbitrary Perturbations

Published in CVPR 2025, 2025

This paper extends AGNNCert to defending posioning attack.

Recommended citation: Jiate Li, Meng Pang, Yun Dong, Binghui Wang. (2025). "Deterministic Certification of Graph Neural Networks against Graph Poisoning Attacks with Arbitrary Perturbations." CVPR 2025.
Download Paper

“Someone Hid It”: Query-Agnostic Black-Box Attacks on LLM-Based Retrieval

Published in ICML 2026, 2026

This paper proposes DQA learning that learn black-box attack against LLMR without rely on specific victim query.

Recommended citation: Jiate Li, Defu Cao, Li Li, Wei Yang, Yuehan Qin, Chenxiao Yu, Tiannuo Yang, Ryan A Rossi, Yan Liu, Xiyang Hu, Yue Zhao. (2026). "“Someone Hid It”: Query-Agnostic Black-Box Attacks on LLM-Based Retrieval." ICML 2026.
Download Paper