Zhecheng Sheng

I am a 5th-year Ph.D. Student in Health Data Science, and a member of Cognitive AI Lab at UMN. I am honored to be advised by Prof. Serguei Pakhomov. I received B.S. in Agricultural Science at Zhejiang University in 2017 (advised by Prof. Jun Zhu) and M.S. in Biostatistics at Duke University in 2019 (advised by Prof. Benjamin Goldstein).

My current research interests include Natural Language Processing and Trustworthy Machine Learning. My on-going projects specifically focus on developing methodologies to ensure fairness and robustness of machine learning and deep learning models and their application in healthcare. This involves addressing the presence of sensitive features or confounding shifts that may interfere with model performances. Besides, I am also interested in designing methods for probing Large Language Models (LLM) to assess the faithfulness of language generation and adapting it for downstream domain-specific tasks.

☕ Feel free to reach out for a quick chat!

news

Aug 20, 2025	One paper accepted to the EMNLP 2025 Main Conference.
Jun 23, 2025	I am starting my summer internship as an Applied Scientist @Amazon. So excited to return to the team and meet everyone again!
Jun 05, 2025	One paper accepted to the Journal of Biomedical Informatics.
May 15, 2025	One paper accepted to ACL 2025 Main Conference! In this paper we propose weight masking methods to mitigate confounding bias introduced during model finetuning.
Apr 16, 2025	I passed my PhD preliminary exam and became a candidate !

selected publications

EMNLP

BBScoreV2: Learning Time-Evolution and Latent Alignment from Stochastic Representation

Tianhao Zhang , Zhecheng Sheng , Zhexiao Lin , and 2 more authors

2025

Bib PDF

@misc{zhang2025bbscorev2learningtimeevolutionlatent,
  title = {BBScoreV2: Learning Time-Evolution and Latent Alignment from Stochastic Representation},
  author = {Zhang, Tianhao and Sheng, Zhecheng and Lin, Zhexiao and Jiang, Chen and Kang, Dongyeop},
  year = {2025},
  journal = {Main Conference of the Empirical Methods in Natural Language Processing (EMNLP) 2025},
  eprint = {2405.17764},
  archiveprefix = {arXiv},
  primaryclass = {cs.CL},
  url = {https://arxiv.org/abs/2405.17764},
}

ACL

Mitigating Confounding in Speech-Based Dementia Detection through Weight Masking

Zhecheng Sheng , Xiruo Ding , Brian Hur , and 3 more authors

2025

Bib PDF

@misc{sheng2025mitigatingconfoundingspeechbaseddementia,
  title = {Mitigating Confounding in Speech-Based Dementia Detection through Weight Masking},
  author = {Sheng, Zhecheng and Ding, Xiruo and Hur, Brian and Li, Changye and Cohen, Trevor and Pakhomov, Serguei},
  year = {2025},
  eprint = {2506.05610},
  journal = {Main Conference of the Association for Computational Linguistics (ACL) 2025},
  archiveprefix = {arXiv},
  primaryclass = {cs.CL},
  url = {https://arxiv.org/abs/2506.05610},
}

JBI

Tailoring task arithmetic to address bias in models trained on multi-institutional datasets

Xiruo Ding , Zhecheng Sheng , Brian Hur , and 5 more authors

Journal of Biomedical Informatics, 2025

Bib

@article{DING2025104858,
  author = {Ding, Xiruo and Sheng, Zhecheng and Hur, Brian and Tauscher, Justin and Ben-Zeev, Dror and Yeti{\c s}gen, Meliha and Pakhomov, Serguei and Cohen, Trevor},
  doi = {https://doi.org/10.1016/j.jbi.2025.104858},
  issn = {1532-0464},
  journal = {Journal of Biomedical Informatics},
  title = {Tailoring task arithmetic to address bias in models trained on multi-institutional datasets},
  url = {https://www.sciencedirect.com/science/article/pii/S1532046425000875},
  volume = {168},
  year = {2025},
  bdsk-url-1 = {https://www.sciencedirect.com/science/article/pii/S1532046425000875},
}

ACL

Too Big to Fail: Larger Language Models are Disproportionately Resilient to Induction of Dementia-Related Linguistic Anomalies

Changye Li , Zhecheng Sheng , Trevor Cohen , and 1 more author

In , Aug 2024

Bib PDF

@inproceedings{li-etal-2024-big,
  title = {Too Big to Fail: Larger Language Models are Disproportionately Resilient to Induction of Dementia-Related Linguistic Anomalies},
  author = {Li, Changye and Sheng, Zhecheng and Cohen, Trevor and Pakhomov, Serguei},
  journal = {Findings of the Association for Computational Linguistics ACL 2024},
  month = aug,
  year = {2024},
  url = {https://aclanthology.org/2024.findings-acl.380/},
}

AAAI

BBScore: A Brownian Bridge Based Metric for Assessing Text Coherence

Zhecheng Sheng , Tianhao Zhang , Chen Jiang , and 1 more author

In , Feb 2024

Bib PDF

@inproceedings{sheng2023bbscore,
  title = {BBScore: A Brownian Bridge Based Metric for Assessing Text Coherence},
  author = {Sheng, Zhecheng and Zhang, Tianhao and Jiang, Chen and Kang, Dongyeop},
  month = feb,
  year = {2024},
  journal = {Proceedings of the 38th Annual AAAI Conference on Artificial Intelligence},
  url = {https://arxiv.org/abs/2312.16893},
}