Gezheng Xu

About Me

Hello! I’m Gezheng, currently a Postdoctoral Research Fellow in the Machine Learning group at the University of Western Ontario. I’m lucky to be co-supervised by Prof. Boyu Wang and Prof. Charles Ling (Fellow, Canadian Academy of Engineering) during my Ph.D. journey from 2021.09 to 2026.02. My research builds from focused technical questions around Transfer Learning, Representation Learning, Algorithmic Fairness, and Uncertainty Estimation, with recent work in time-series foundation models and data protection. Scaling outward, I am interested in how these foundations apply to broader challenges: distribution shift and uncertainty in multimodal learning, knowledge transfer in LLMs and agentic systems, and AI for Science. At the largest scale, I care deeply about AI safety and privacy, and about how advances in AI can translate into practical impact across science, industry, and everyday life.

Before starting my Ph.D., I earned my Bachelor’s degree in Mathematics and Applied Mathematics from Beihang University in 2018. Later, I completed dual master’s degrees at Beihang University and CentraleSupélec (Paris-Saclay University) in 2021. During this time, I conducted research in biomedical NLP under the guidance of Prof. Wenge Rong, a formative experience that shaped my early research mindset and sparked my passion for building models that generalize across domains.

I’m currently on the job market, seeking postdoctoral or faculty positions. If any of my work resonates with you, whether it’s a research discussion, a collaboration, or a potential opportunity, please feel free to reach out!

Recent Activities

Jan, 2025	Our paper, “Unraveling the Mysteries of Label Noise in Source-Free Domain Adaptation: Theory and Practice.”, has been accepted to IEEE TPAMI. Congratulations to Li!
Jan, 2025	Our papers, “Revisiting Source-Free Domain Adaptation: a New Perspective via Uncertainty Control” and “ZETA: Leveraging Z-order Curves for Efficient Top-k Attention.”, have been accepted at ICLR 2025. Congratulations to Hui and Qiuhao!
Dec, 2024	Our paper, Leveraging Group Classification with Descending Soft Labeling for Deep Imbalanced Regression, has been accepted for an oral presentation at AAAI 2025. Congratulations to Ruizhi!
May, 2024	Our paper, “Intersectional Unfairness Discovery”, was accepted at ICML 2024.
Dec, 2023	Our paper, “Generalizing across Temporal Domains with Koopman Operators”, was accepted at AAAI 2024.
May, 2023	Our paper, “Label Shift Conditioned Hybrid Querying for Deep Active Learning”, has been accepted at Knowledge-Based Systems.
May, 2023	Attending the ICLR conference onsite in Rwanda and delivered a spotlight presentation!
Apr, 2023	Our paper, “Towards More General Loss and Setting in Unsupervised Domain Adaptation”, was accepted at IEEE Transactions on Knowledge and Data Engineering (TKDE).
Jan, 2023	Our paper, “When Source-Free Domain Adaptation Meets Learning with Noisy Labels”, was accepted at ICLR 2024.
Jan, 2023	Our paper, “Gap Minimization for Knowledge Sharing and Transfer”, was accepted at Journal of Machine Learning Research (JMLR).
Sep, 2022	Our paper, “On Learning Fairness and Accuracy on Multiple Subgroups”, was accepted at NeurIPS 2022.
Sep, 2021	Begin PhD studies at University of Western Ontario (Western University).
May, 2021	Our paper, “External Features Enriched Model for Biomedical Question Answering”, was accepted at BMC Bioinformatics.

Selected Publications

(Note: * indicates equal contribution.)

TPAMI

Unraveling the Mysteries of Label Noise in Source-Free Domain Adaptation: Theory and Practice

Gezheng Xu^*, Li Yi^*, Pengcheng Xu, Jiaqi Li, Ruizhi Pu, Changjian Shui, Charles Ling, A. Ian McLeod, and Boyu Wang

IEEE Transactions on Pattern Analysis and Machine Intelligence, 2025

DOI Bib Code

@article{xu2025unraveling,
  author = {Xu, Gezheng and Yi, Li and Xu, Pengcheng and Li, Jiaqi and Pu, Ruizhi and Shui, Changjian and Ling, Charles and McLeod, A. Ian and Wang, Boyu},
  journal = {IEEE Transactions on Pattern Analysis and Machine Intelligence},
  title = {Unraveling the Mysteries of Label Noise in Source-Free Domain Adaptation: Theory and Practice},
  year = {2025},
  doi = {10.1109/TPAMI.2025.3536845},
}

ICLR

Revisiting Source-Free Domain Adaptation: a New Perspective via Uncertainty Control

Gezheng Xu^*, Hui Guo^*, Li Yi, Charles Ling, Boyu Wang, and Grace Yi

In International Conference on Learning Representations (ICLR), 2025

Bib Code

@inproceedings{xu2025revisiting,
  title = {Revisiting Source-Free Domain Adaptation: a New Perspective via Uncertainty Control},
  author = {Xu, Gezheng and Guo, Hui and Yi, Li and Ling, Charles and Wang, Boyu and Yi, Grace},
  booktitle = {International Conference on Learning Representations (ICLR)},
  year = {2025},
}

ICML

Intersectional Unfairness Discovery

Gezheng Xu, Qi Chen, Charles Ling, Boyu Wang, and Changjian Shui

In International Conference on Machine Learning (ICML), 2024

Bib HTML PDF Code Poster

@inproceedings{pmlr-v235-xu24d,
  title = {Intersectional Unfairness Discovery},
  author = {Xu, Gezheng and Chen, Qi and Ling, Charles and Wang, Boyu and Shui, Changjian},
  booktitle = {International Conference on Machine Learning (ICML)},
  pages = {54888--54917},
  year = {2024},
  editor = {Salakhutdinov, Ruslan and Kolter, Zico and Heller, Katherine and Weller, Adrian and Oliver, Nuria and Scarlett, Jonathan and Berkenkamp, Felix},
  volume = {235},
  series = {Proceedings of Machine Learning Research},
  publisher = {PMLR},
}

ICLR

When Source-Free Domain Adaptation Meets Learning with Noisy Labels

Li Yi^*, Gezheng Xu^*, Pengcheng Xu, Jiaqi Li, Ruizhi Pu, Charles Ling, A Ian McLeod, and Boyu Wang

International Conference on Learning Representations (ICLR), 2023

Bib Video Code Slides

@article{yi2023source,
  title = {When Source-Free Domain Adaptation Meets Learning with Noisy Labels},
  author = {Yi, Li and Xu, Gezheng and Xu, Pengcheng and Li, Jiaqi and Pu, Ruizhi and Ling, Charles and McLeod, A Ian and Wang, Boyu},
  journal = {International Conference on Learning Representations (ICLR)},
  year = {2023},
}

NeurIPS

On Learning Fairness and Accuracy on Multiple Subgroups

Changjian Shui^*, Gezheng Xu^*, Qi Chen, Jiaqi Li, Charles X Ling, Tal Arbel, Boyu Wang, and Christian Gagné

Advances in Neural Information Processing Systems (NeurIPS), 2022

Bib HTML Code

@article{shui2022learning,
  title = {On Learning Fairness and Accuracy on Multiple Subgroups},
  author = {Shui, Changjian and Xu, Gezheng and Chen, Qi and Li, Jiaqi and Ling, Charles X and Arbel, Tal and Wang, Boyu and Gagn{\'e}, Christian},
  journal = {Advances in Neural Information Processing Systems (NeurIPS)},
  volume = {35},
  pages = {34121--34135},
  year = {2022},
}

BMC Bioinformatics

External Features Enriched Model for Biomedical Question Answering

Gezheng Xu, Wenge Rong, Yanmeng Wang, Yuanxin Ouyang, and Zhang Xiong

BMC Bioinformatics, 2021

Bib Code

@article{xu2021external,
  title = {External Features Enriched Model for Biomedical Question Answering},
  author = {Xu, Gezheng and Rong, Wenge and Wang, Yanmeng and Ouyang, Yuanxin and Xiong, Zhang},
  journal = {BMC Bioinformatics},
  volume = {22},
  number = {1},
  pages = {272},
  year = {2021},
  publisher = {Springer},
}