Zachary Chase, Kent State University Postdoc：Optimal Mistake Bounds for Transductive Online Learning

Описание: “Best Paper Deep Dive” is a series of academic events hosted by Chaspark, focusing on in-depth interpretation and discussion of high-quality research papers. We invite authors of award-winning papers from top conferences to deliver Live Presentation, in which they provide in-depth explanations of their work, including research background, key innovations, and technical details. You can discuss academic ideas directly with paper authors and learn about the latest cutting-edge research achievements.

At 19:00 on February 26, Zachary Chase, recipient of the NeurIPS 2025 Best Paper Runner-up Award and Kent State University Postdoc, will join the Best Paper Deep Dive Vol 44. He will share insights based on his award-winning paper published at NeurIPS 2025, a top conference in AI, focusing on the topic “Optimal Mistake Bounds for Transductive Online Learning”.

Abstract: It is well-known that the Littlestone dimension characterizes the optimal mistake bound for online learning in the realizable setting with a hypothesis class. But what if the learner can see the sequence in advance; how much better can they do with access to this "unlabelled data"? We provide the tight answer to this question.

Bio: Zachary Chase completed his PhD at the University of Oxford in combinatorial number theory. He has done postdocs at the Technion, UCSD, and Kent State University. He won the STOC 2021 best student paper award as well as NeurIPS 2025 best paper runner up.

References
1. Zachary Chase, Steve Hanneke, Shay Moran, Jonathan Shafer ,“ Optimal Mistake Bounds for Transductive Online Learning, ” 39th Conference on Neural Information Processing Systems (NeurIPS 2025 Best Paper Runner-up)