Hi there 
Welcome to my Homepage!
Hi! I’m Xinyu Yan (Tristan), currently a Ph.D. student at College of Computing and Data Science, Nanyang Technological University, where I am fortunate to be supervised by Prof. Wei Yang Bryan Lim. I am also affiliated with the Alibaba-NTU Global e-Sustainability CorpLab (ANGEL).
My research interests primarily focus on Trustworthy AI (LLM Unlearning, Deepfake Detection) and Efficient AI (Efficient LLM Serving, Edge Computing), as well as other emerging areas in AI. I enjoy exploring diverse research directions and collaborating with researchers across different fields.
Outside of research, I have a deep passion for History and Philosophy, which often inspire the way I think about problems. I am also an avid fan of Anime and enjoy Traveling and Hiking whenever I get the chance 🏔️, as there is something about reaching a summit that feels not too different from finishing a paper.
Please feel free to reach out 🥳, if you are interested in collaboration or potential opportunities!
News
- 2026.01 🎓 I receive my M.Sc. in Artificial Intelligence from Nanyang Technological University (NTU).
- 2026.01 🎉 One paper has been accepted at ICLR 2026 (Poster).
- 2026.01 🌟 I am also affiliated with the Alibaba-NTU Global e-Sustainability CorpLab (ANGEL).
- 2026.01 🥰 I begin my Ph.D. studies at Nanyang Technological University (NTU).
- 2026.01 🎉 One paper has been accepted at EACL 2026 Main Conference (Oral).
- 2025.11 🎉 One paper has been accepted at AAAI-26 Main Technical Track (Poster).
- 2025.09 🎉 One paper has been accepted at NeurIPS 2025 (Poster).
- 2024.08 🥳 I join the Midori Lab at Nanyang Technological University as a Research Assistant.
- 2024.08 🥰 I begin my master's studies at Nanyang Technological University (NTU).
- 2024.07 🎓 I receive my B.Eng. in Computer Science and Technology from Shanghai University (SHU).
Experience
Nanyang Technological University
2026.01 - Present
Ph.D. at College of Computing and Data Science, supervised by Prof. Wei Yang Bryan Lim
2026.01 - Present
Ph.D. at College of Computing and Data Science, supervised by Prof. Wei Yang Bryan Lim
Nanyang Technological University
2024.08 - 2026.01
M.Sc. in Artificial Intelligence at College of Computing and Data Science
2024.08 - 2026.01
M.Sc. in Artificial Intelligence at College of Computing and Data Science
Shanghai University
2020.09 - 2024.07
B.Eng. in Computer Science and Technology at School of Computer Engineering and Science
2020.09 - 2024.07
B.Eng. in Computer Science and Technology at School of Computer Engineering and Science
Publications
(* Equal Contribution · † Corresponding Author · ‡ Project Leader)
FraudBench: A Multimodal Benchmark for Detecting AI-Generated Fraudulent Refund Evidence
Xinyu Yan, Boyang Chen, Jiaming Zhang, Tiantong Wu, Hong Xi Tae,
Yichen He, Tiantong Wang, Yachun Mi, Yurong Hao, Yilei Zhao,
Lei Xiao, Longtao Huang, Pengjun Xie, Wei Liu, Wei Yang Bryan Lim.
We introduce FraudBench, the first multimodal benchmark for detecting AI-generated fraudulent refund evidence.
Underview [Paper] [Project Page] [Code] [Dataset]
Xinyu Yan, Boyang Chen, Jiaming Zhang, Tiantong Wu, Hong Xi Tae,
Yichen He, Tiantong Wang, Yachun Mi, Yurong Hao, Yilei Zhao,
Lei Xiao, Longtao Huang, Pengjun Xie, Wei Liu, Wei Yang Bryan Lim.
We introduce FraudBench, the first multimodal benchmark for detecting AI-generated fraudulent refund evidence.
Underview [Paper] [Project Page] [Code] [Dataset]
MPU: Towards Secure and Privacy-Preserving Knowledge Unlearning for Large Language Models
Tiantong Wang, Xinyu Yan, Tiantong Wu†, Yurong Hao, Pengjun Xie, Wei Yang Bryan Lim.
We present MPU, an algorithm-agnostic framework enabling secure machine unlearning for LLMs under strict dual non-disclosure constraints, without sharing server parameters or client data.
Underview [Paper] [Code]
Tiantong Wang, Xinyu Yan, Tiantong Wu†, Yurong Hao, Pengjun Xie, Wei Yang Bryan Lim.
We present MPU, an algorithm-agnostic framework enabling secure machine unlearning for LLMs under strict dual non-disclosure constraints, without sharing server parameters or client data.
Underview [Paper] [Code]
Oblivionis: A Lightweight Learning and Unlearning Framework for Federated Large Language Models
Fuyao Zhang, Xinyu Yan, Tiantong Wu, Wenjie Li, Tianxiang Chen,
Yang Cao, Ran Yan, Longtao Huang, Wei Yang Bryan Lim†, Qiang Yang.
We introduce Oblivionis, a lightweight framework that integrates federated learning and targeted unlearning for LLMs, to enable privacy-preserving training and compliance with GDPR's right to be forgotten.
AAAI-26 [Paper] [Project Page] [Code]
Fuyao Zhang, Xinyu Yan, Tiantong Wu, Wenjie Li, Tianxiang Chen,
Yang Cao, Ran Yan, Longtao Huang, Wei Yang Bryan Lim†, Qiang Yang.
We introduce Oblivionis, a lightweight framework that integrates federated learning and targeted unlearning for LLMs, to enable privacy-preserving training and compliance with GDPR's right to be forgotten.
AAAI-26 [Paper] [Project Page] [Code]
- Underview FraudBench: A Multimodal Benchmark for Detecting AI-Generated Fraudulent Refund Evidence
- Underview MPU: Towards Secure and Privacy-Preserving Knowledge Unlearning for Large Language Models
- ICLR 2026 AdaCache: Adaptive Caching and Context Augmentation for Efficient LLM Serving
Poster. - EACL 2026 SEMIROUTER: Sparse-Data Enhanced Routing for Adaptive Multi-LLM System
Oral. - AAAI 2026 Oblivionis: A Lightweight Learning and Unlearning Framework for Federated Large Language Models
Poster. - NeurIPS 2025 FedRAM: Federated Reweighting and Aggregation for Multi-Task Learning
Poster.
Awards
- 2026.05, ICML 2026 Gold Reviewer Award.
- 2025.11, AAAI-26 Scholarship & Volunteer Award.
Services
- Technical Reviewer for ICML 2026 (Gold Reviewer).
- Reviewer for NeurIPS 2026 Evaluations and Datasets Track.
- Reviewer for the Transactions on Machine Learning Research (TMLR)
Talks
- 2025.05, Efficient LLM Serving, Institute of Science Tokyo, Tokyo, Japan.
Mottos
“Man’s dearest possession is life. It is given to him but once, and he must live it so as to feel no torturing regrets for wasted years, never know the burning shame of a mean and petty past; so live that, dying he might say: all my life, all my strength were given to the finest cause in all the world—the fight for the Liberation of Mankind.”
— Nikolai Ostrovsky, How the Steel Was Tempered
“Trust the fact that history will judge events honestly and you will never have to be embarrassed for what your father did. On no account ever be one of those people who criticize but do not follow through their actions. Such people are hypocrites, weak, worthless people who do not have the power to reconcile their beliefs with their actions. I wish you courage, my dear. Be strong in the belief that life is wonderful. Be positive and believe that the Revolution will always win.”
— Sablin's last letter to his son before his execution