Welcome to Friends of Optimization, a seminar series featuring one-hour talks on the latest advancements in optimization. The seminar series is mostly virtual and is organized by Shuvomoy Das Gupta (Rice CMOR) and Ernest K. Ryu (OpenAI, UCLA Math). 

Upcoming talk

• Disciplined Nonlinear Programming by Daniel Cederberg (March 24, Tuesday)

Past talks

We started Friends of Optimization in December 2024. These links below also contain some past talks that we organized.

• Negative Stepsizes Make Gradient-Descent-Ascent Converge by Henry Shugart, Aug 29 (virtual) [Video link]

• Provable Non-Accelerations of the Heavy-Ball Method by Aymeric Dieuleveut, March 10, 2025 at Columbia (in person) [Video link] 

• Algebraic Methods in Convex Optimization by Kevin Shu, February 13, 2025 at UCLA (in person) [Video link]

• How to Make the Gradient Descent-Ascent Converge to Local Minima by Donghwan Kim, January 08, 2025 at UCLA (in person) [Video link]

• Schedules & Schedule-Free Learning by Aaron Defazio, December 6, 2024 at UCLA (in person) [Video link] 

• Acceleration by Stepsize Hedging by Jason Altschuler, December 4, 2023 at MIT (in person) [Video link] 

• Provably Faster Gradient Descent via Long Steps by Ben Grimmer, August 11, 2023, MIT (virtual) [Video link] 

• Parameter-Free Adaptive Methods for Deep Learning by Konstantin Mishchenko, June 26, 2023, MIT (virtual) [Video link]