EE Seminar: The Price of Adaptivity in Stochastic Convex Optimization
הרישום לסמינר יבוצע באמצעות סריקת הברקוד למודל (יש להיכנס לפני כן למודל, לא באמצעות האפליקציה) - )- הרישום מסתיים ב- 13:10
Registration to the seminar will be done by scanning the barcode for the Moodle (Please enter ahead to the Moodle, NOT by application)- Registration ends at 13:10
(The talk will be given in English)
Speaker: Dr. Yair Carmon
Blavatnik School of Computer Science and AI, Tel Aviv University
|
011 hall, Electrical Engineering-Kitot Building |
Monday, June 9th, 2025
13:00 - 14:00
|
|
The Price of Adaptivity in Stochastic Convex Optimization
Abstract
While stochastic optimization methods drive continual improvements in machine learning, choosing the optimization parameters—particularly the learning rate (LR)—remains challenging. In this talk, I will present our work on designing tuning-free algorithms, and characterizing the fundamental costs of not knowing key problem parameters in advance. Inspired by the Price of Anarchy in algorithmic game theory, we define the Price of Adaptivity which measures the multiplicative performance drop due to uncertainty in problem parameters, and - in a number of settings - establish nearly matching upper and lower bounds.
The talk is based on joint work with Maor Ivgi, Itai Kreisler, and Oliver Hinder.
Short Bio
Yair Carmon is an assistant professor of computer science at Tel Aviv university (currently on leave). He works on the foundations of optimization and machine learning, focusing on questions about fundamental limits and robustness. Yair received a PhD from Stanford University, advised by John Duchi and Aaron Sidford, and M.Sc. and B.Sc. degrees from the Technion.
