Sitemap
A list of all the posts and pages found on the site. For you robots out there, there is an XML version available for digesting as well.
Pages
Posts
publications
Sparse Adversarial Attack for Video via Gradient-Based Keyframe Selection Permalink
Published in ICASSP 2022, 2022
Official publication: IEEE.
Recommended citation: Yixiao Xu, Xiaolei Liu, Mingyong Yin, Teng Hu, and Kangyi Ding. (2022). "Sparse Adversarial Attack for Video via Gradient-Based Keyframe Selection." ICASSP 2022, 2874–2878.
Efficient Intrusion Detection toward IoT Networks Using Cloud–Edge Collaboration Permalink
Published in Computer Networks, 2023
Official publication: Computer Networks.
Recommended citation: Run Yang, Hui He, Yixiao Xu, Bangzhou Xin, Yulong Wang, Yue Qu, and Weizhe Zhang. (2023). "Efficient Intrusion Detection toward IoT Networks Using Cloud–Edge Collaboration." Computer Networks, 228, 109724.
Sparse Black-Box Inversion Attack with Limited Information Permalink
Published in ICASSP 2023, 2023
Official publication: IEEE.
Recommended citation: Yixiao Xu, Xiaolei Liu, Teng Hu, Bangzhou Xin, and Run Yang. (2023). "Sparse Black-Box Inversion Attack with Limited Information." ICASSP 2023, 1–5.
Watch Your Words: Successfully Jailbreak LLM by Mitigating the Prompt Malice Permalink
Published in APWeb-WAIM 2024, 2024
Official publication: Springer.
Recommended citation: Xiaowei Xu, Yixiao Xu, Xiong Chen, Peng Chen, Mohan Li, and Yanbin Sun. (2024). "Watch Your Words: Successfully Jailbreak LLM by Mitigating the Prompt Malice." Web and Big Data, Lecture Notes in Computer Science 14961, 295–309.
LT-Defense: Searching-free Backdoor Defense via Exploiting the Long-tailed Effect Permalink
Published in NeurIPS 2024, 2024
Official publication: NeurIPS Proceedings.
Recommended citation: Yixiao Xu, Binxing Fang, Mohan Li, Keke Tang, and Zhihong Tian. (2024). "LT-Defense: Searching-free Backdoor Defense via Exploiting the Long-tailed Effect." Advances in Neural Information Processing Systems, 37, 3543–3563.
A Lightweight Defense Against Jailbreak Attacks via Honeypot Prompts Permalink
Published in AANN 2025, 2025
Best Paper Award.
Recommended citation: Ximin Zheng, Yixiao Xu, and Mohan Li. (2025). "A Lightweight Defense Against Jailbreak Attacks via Honeypot Prompts." 2025 5th International Conference on Advanced Algorithms and Neural Networks, 347–355.
Query-Efficient Model Inversion Attacks: An Information Flow View Permalink
Published in IEEE Transactions on Information Forensics and Security, 2025
Official publication: IEEE. Published online on 18 December 2024.
Recommended citation: Yixiao Xu, Binxing Fang, Mohan Li, Xiaolei Liu, and Zhihong Tian. (2025). "Query-Efficient Model Inversion Attacks: An Information Flow View." IEEE Transactions on Information Forensics and Security, 20, 1023–1036.
Neural Honeypoint: An Active Defense Framework Against Model Inversion Attacks Permalink
Published in IEEE Transactions on Neural Networks and Learning Systems, 2025
Official publication: IEEE Xplore. DOI: 10.1109/TNNLS.2025.3554217.
Recommended citation: Yixiao Xu, Mohan Li, Binxing Fang, Yuan Liu, and Zhihong Tian. (2025). "Neural Honeypoint: An Active Defense Framework Against Model Inversion Attacks." IEEE Transactions on Neural Networks and Learning Systems, 36(9), 16186–16197.
Backdoor-vaccine: An Active Backdoor Defense Framework via Competitive Backdoor Injection Permalink
Published in Neurocomputing, 2026
Official publication: Neurocomputing.
Recommended citation: Feng Zhang, Yixiao Xu, Jun Lv, Zexin Yu, Luo Zhang, Jun Feng, and Xiaolei Liu. (2026). "Backdoor-vaccine: An Active Backdoor Defense Framework via Competitive Backdoor Injection." Neurocomputing, 664, 132133.
AT-Field: Rethinking the Games in Adversarial Training Permalink
Published in AAAI 2026, 2026
Oral presentation.
Recommended citation: Yixiao Xu, Mohan Li, Zhijie Shen, Yuan Liu, and Zhihong Tian. (2026). "AT-Field: Rethinking the Games in Adversarial Training." Proceedings of the AAAI Conference on Artificial Intelligence, 40(32), 27368–27376.
LoopLLM: Transferable Energy-Latency Attacks in LLMs via Repetitive Generation Permalink
Published in AAAI 2026, 2026
Official publication: AAAI. DOI: 10.1609/aaai.v40i38.40445.
Recommended citation: Xingyu Li, Xiaolei Liu, Cheng Liu, Yixiao Xu, Kangyi Ding, Bangzhou Xin, and Jia-Li Yin. (2026). "LoopLLM: Transferable Energy-Latency Attacks in LLMs via Repetitive Generation." Proceedings of the AAAI Conference on Artificial Intelligence, 40(38), 31770–31777.
Artificial Intelligence Resilience: Current State and Future Perspectives Permalink
Published in Strategic Study of CAE, 2026
Official publication: Strategic Study of CAE. DOI: 10.15302/J-SSCAE-2025.08.022.
Recommended citation: Mohan Li, Yixiao Xu, Yanbin Sun, Zhihong Tian, and Binxing Fang. (2026). "Artificial Intelligence Resilience: Current State and Future Perspectives." Strategic Study of CAE, 28(2), 26–42.
BDpackets: A Clean-label Backdoor Attack on Network Traffic Classifiers via Feature Fusion Permalink
Published in IEEE INFOCOM 2026, 2026
Bibliographic record: DBLP.
Recommended citation: Mengxia Zhang, Yixiao Xu, Mohan Li, Yanbin Sun, Meng Han, and Zhihong Tian. (2026). "BDpackets: A Clean-label Backdoor Attack on Network Traffic Classifiers via Feature Fusion." IEEE INFOCOM 2026, 1–10.
Neural Honeytrace: Plug&Play Watermarking Framework against Model Extraction Attacks Permalink
Accepted at ICML 2026, 2026
Accepted as a poster at ICML 2026. OpenReview.
Recommended citation: Yixiao Xu, Binxing Fang, Rui Wang, Yinghai Zhou, Yuan Liu, Mohan Li, and Zhihong Tian. (2026). "Neural Honeytrace: Plug&Play Watermarking Framework against Model Extraction Attacks." Proceedings of the 43rd International Conference on Machine Learning. Accepted, Poster.
NEO: Navigating Entropy in Optimized Black-Box Video Adversarial Attacks
Accepted at IEEE TDSC, 2026
Accepted by IEEE Transactions on Dependable and Secure Computing (TDSC) in July 2026.
Recommended citation: Xiaolei Liu, Yixiao Xu*, Bangzhou Xin, Kangyi Ding, Jia-Li Yin, Shouling Ji, and Xiaosong Zhang. (2026). "NEO: Navigating Entropy in Optimized Black-Box Video Adversarial Attacks." IEEE Transactions on Dependable and Secure Computing. Accepted.
