Most Significant Papers
Prevalence of Neural Collapse During the Terminal Phase of Deep Learning Training
Vardan Papyan*, X.Y. Han*, and David L. Donoho
(*Equal Contribution. Non-alphabetical to balance visibility in citations.)
Proceedings of the National Academy of Sciences (PNAS), 117.40 (2020): 24652-24663.
🌟 Discovered neural collapse, now a widely studied phenomenon in AI training.
Neural Collapse Under MSE Loss: Proximity to and Dynamics on the Central Path
X.Y. Han*, Vardan Papyan*, and David L. Donoho
(*Equal Contribution)
International Conference on Learning Representations (ICLR) 2022, 26 April 2022. (Oral)
🏆 ICLR 2022 Outstanding Paper Award
Most Recent Work(s)
A Theoretical Framework for Auxiliary-Loss-Free Load Balancing of Sparse Mixture-of-Experts in Large-Scale AI Models
X.Y. Han* and Yuan Zhong*
NeurIPS 2025, MLxOR Workshop (arXiv:2512.03915)
💬 What can operations contribute to AI? This paper is my first answer. s-MoE load-balancing procedures are examples of ML heuristics that work very well in practice but we don’t have much rigorous understanding for why they would. In this paper, Yuan and I use mathematical tools from OR/OM to build a theoretical framework for analyzing why DeepSeek’s ALF-LB procedure is effective at s-MoE load-balancing.
Peer-Reviewed Publications
Survey Descent: A Multipoint Generalization of Gradient Descent for Nonsmooth Optimization
X.Y. Han and Adrian S. Lewis
SIAM Journal on Optimization (SIOPT) 2023, 33(1), 36-62.
🏆 ICCOPT 2022 Best Paper Prize for Young Researchers (Finalist)
Follow-Up: See Initialization of Survey Descent (Chapter 4) in my PhD Dissertation.
Artificial Intelligence and Discovering the Digitised Photoarchive (Book Chapter)
X.Y. Han*, Vardan Papyan*, Ellen Prokop*, David L. Donoho, C. Richard Johnson Jr.
(*Equal Contribution)
Archives, Access and AI: Working with Born-Digital and Digitised Archival Collections, 2022. Editor: Lise Jaillant. Part of Digital Humanities Research Series. Bielefeld University Press.
Expert Surgeons and Deep Learning Models Can Predict the Outcome of Surgical Hemorrhage From 1 Min of Video
Dhiraj J Pangal, Guillaume Kugener, Yichao Zhu, Aditya Sinha, Vyom Unadkat, David J Cote, Ben Strickland, Martin Rutkowski, Andrew Hung, Animashree Anandkumar, X.Y. Han, Vardan Papyan, Bozena Wrobel, Gabriel Zada, Daniel A Donoho
Scientific Reports 12.1 (2022): 1-10, 12 May 2022
Neural Collapse Under MSE Loss: Proximity to and Dynamics on the Central Path
X.Y. Han*, Vardan Papyan*, and David L. Donoho
(*Equal Contribution)
International Conference on Learning Representations (ICLR) 2022, 26 April 2022. (Oral)
🏆 ICLR 2022 Outstanding Paper Award
Short Talks: Oral [15 min] Poster [5 min]
Full-Length Talk: Talk [60 mins]
Automatic Assessment of Surgical Performance Using Intraoperative Video and Deep Learning: A Comparison with Expert Surgeon Video Review
Dhiraj J Pangal, Guillaume Kugener, Yichao Zhu, Aditya Sinha, Vyom Unadkat, David J Cote, Arman Roshannai, Ben Strickland, Martin Rutkowski, Andrew Hung, Animashree Anandkumar, X.Y. Han, Vardan Papyan, Bozena Wrobel, Gabriel Zada, Daniel A Donoho
Journal of Neurological Surgery Part B: Skull Base, 83(S 01), A049, 15 February 2022
Utility of the Simulated Outcomes Following Carotid Artery Laceration Video Data Set for Machine Learning Applications
Dhiraj Pangal, Guillaume Kugener, Yichao Zhu, Aditya Sinha, Vyom Unadkat, David Cote, Ben Strickland, Martin Rutkowski, Andrew Hung, Animashree Anandkumar, X.Y. Han, Vardan Papyan, Bozena Wrobel, Gabriel Zada, Daniel A Donoho
JAMA Network Open, 5(3), e223177-e223177, 31 January 2022
AI and the Digitized Photoarchive
Ellen Prokop*, X.Y. Han*, Vardan Papyan*, David L. Donoho, C. Richard Johnson Jr
(*Equal Contribution. Non-alphabetical to balance visibility in citations.)
Art Documentation, 40.1 (2021): 1-20.
Prevalence of Neural Collapse During the Terminal Phase of Deep Learning Training
Vardan Papyan*, X.Y. Han*, and David L. Donoho
(*Equal Contribution. Non-alphabetical to balance visibility in citations.)
Proceedings of the National Academy of Sciences (PNAS), 117.40 (2020): 24652-24663.
Talk [60 mins]
Sparse Median Graphs Estimation in a High Dimensional Semiparametric Model
Fang Han, Xiaoyan Han, Han Liu, and Brian Caffo
The Annals of Applied Statistics, 10.3 (2016): 1397-1426.
Pre-Prints
Disk Matrices and the Proximal Mapping for the Numerical Radius
X.Y. Han and Adrian S. Lewis
Arxiv, 30 April 2020
Talk by Adrian Lewis (6:18-13:50)
Software and Repositories
autoNSO: Autodifferentiation-based framework in Python for benchmarking common nonsmooth optimization methods. https://github.com/xiaoyanh/autoNSO.
neuralcollapse: Self-contained Pytorch code for reproducing the Neural Collapse phenomenon. https://github.com/neuralcollapse/neuralcollapse.
X.Y. Han 