Data-Intensive Distributed Systems LaboratoryCS Seminar
Date: March 25th, 2026
Time: 12:50pm
Room: SB111
Dr. Yutong Wang
Assistant Professor
Computer Science Department
Illinois Institute of Technology
Talk Title
Gradient-as-retrieval: Classification beyond Cross Entropy
Talk Abstract
Cross entropy (CE) is the loss of choice for classification tasks. However, computing the CE loss and gradient requires transcendental functions which can become expensive. The transcendental function-free familywise (FW) loss has been shown to enjoy strictly better statistical guarantees than the CE loss. In this work, we prove theoretical results that enable efficient computation of the gradient of the FW loss using "retrieval-style" algorithms. Based on our theory, we provide practical implementations. A challenge in designing new loss functions is that widely adopted optimizers and learning rate schedules are tuned to CE. Experimentally, we demonstrate that the FW loss outperforms cross entropy when using parameter-free learning methods.
Speaker Bio
Yutong Wang is an Assistant Professor in the Computer Science Department at Illinois Institute of Technology, where he leads the EMC2 (Elements of MultiClass Classification) Lab. He is also a member of the IDEAL Institute. His research focuses on the theoretical and practical aspects of classification, including uncertainty estimation, calibration, and privacy-preserving machine learning. He is the recipient of an NSF CRII award on uncertainty estimation and robustness in hierarchical classification. Previously, Yutong was a Postdoctoral Research Fellow supported by the Eric and Wendy Schmidt AI Fellowship in at the University of Michigan, Ann Arbor. He completed his Ph.D. in Electrical and Computer Engineering at the University of Michigan and obtained his Masters in Mathematics at the University of California, Davis.