Jiali Cheng

I am always open to collaboration, whether you are a seniro / junior / beginner researcher.

I am a PhD student from UMass Lowell, advised by Prof. Hadi Amiri. Previously

Intern at Amazon Alexa AI (2026 & 2025), advised by Dr. Oleg Rokhlenko, Dr. Xingzhi Guo, Anjishnu Kumar
RA at Harvard Medical School, advised by Prof. Marinka Zitnik
RA at THUNLP, advised by Prof. Zhiyuan Liu, Dr. Yusheng Su

Research

My research aims to build Trustworthy AI Systems and advance AI Applications, focusing on three main directions:

Trustworthy of Controllable AI.
- Knowledge Control. (Targeted Unlearning in GNN, VLM, Tool LLM, GenRec, Cross-Modal Benchmark)
- Knowledge Conflicts. I study how conflicts arise when agents interact with environment, and how to resolve them (Tool-Memory Conflict).
- Robustness (FairFlow, Robustness under Distillation)
- Interpretability (Mode Connectivity in Unlearning, Circuit-Guided Unlearning Difficulty)
AI for Science and Applications.
- Medical and Healthcare. (MedDec, Demential Detection)
- Psychology. (Human Emotions)

I am looking for full-time industry / academic opportunities. Please feel free to contact me if you think I’m a good fit.

News

Jul 08, 2026	🎉🚀1 paper accepted to COLM 2026: Mode Connectivity in Unlearning! This is the first loss landscape analysis of machine unlearning.
May 01, 2025	🔥🔥ToolDelete is accepted to ICML 2025! This is the first work on unlearning tools from tool-augmented LLMs.
Apr 28, 2025	🔥🔥I was awarded the Computer Science Outstanding Graduate Research Award for Ph.D. Students at UMass Lowell! Many thanks to my advisor Prof. Amiri and collaborators!
Apr 01, 2025	🔥🔥Two preprints available: Exploring Mode Connectivity in the context of Unlearning, and Fair Graph Unlearning.
Jan 22, 2025	🔥🔥Two papers accepted to NAACL 2025! Main Conference (Short): EqualizeIR: Mitigating Linguistic Biases in Retrieval Models, and Workshop on Cognitive Modeling and Computational Linguistics (CMCL) Workshop: Linguistic Blind Spots of Large Language Models.
Oct 18, 2024	🔥🔥 MU-Bench is accepted to SafeGenAI @ NeurIPS’24! Stay tuned for code and leaderboard.

Selected Publications

Trustworthy

Tool Unlearning for Tool-Augmented LLMs

Jiali Cheng, and Hadi Amiri

In ICML , 2025

Bib

@inproceedings{cheng2025toom,
  title = {Tool Unlearning for Tool-Augmented LLMs},
  author = {Cheng, Jiali and Amiri, Hadi},
  year = {2025},
  booktitle = {ICML},
  url = {https://arxiv.org/abs/2502.01083}
}

Trustworthy

MU-Bench: A multitask multimodal benchmark for machine unlearning

Jiali Cheng, and Hadi Amiri

In NeurIPS SafeGenAI Workshop , 2024

arXiv Bib

@inproceedings{cheng2024mu,
  title = {MU-Bench: A multitask multimodal benchmark for machine unlearning},
  author = {Cheng, Jiali and Amiri, Hadi},
  booktitle = {NeurIPS},
  year = {2024},
  url = {https://arxiv.org/abs/2406.14796}
}

Trustworthy

MultiDelete for Multimodal Machine Unlearning

Jiali Cheng, and Hadi Amiri

In ECCV , 2024

arXiv Bib

@inproceedings{cheng2024multidelete,
  title = {MultiDelete for Multimodal Machine Unlearning},
  author = {Cheng, Jiali and Amiri, Hadi},
  booktitle = {ECCV},
  year = {2024},
  url = {https://arxiv.org/abs/2311.12047}
}

Trustworthy

GNNDelete: A General Strategy for Unlearning in Graph Neural Networks

Jiali Cheng, George Dasoulas, Huan He, Chirag Agarwal, and Marinka Zitnik

In ICLR , 2023

arXiv Bib

@inproceedings{cheng2023gnndelete,
  title = {{GNND}elete: A General Strategy for Unlearning in Graph Neural Networks},
  author = {Cheng, Jiali and Dasoulas, George and He, Huan and Agarwal, Chirag and Zitnik, Marinka},
  booktitle = {ICLR},
  year = {2023},
  url = {https://arxiv.org/abs/2302.13406}
}