My sessions at ICCCOPT

I was at the ICCOPT conference in Berlin from 5 to 8 August as the chair of 3 sessions: 2 on variational inequality/minimax/GANs and 1 on non-smooth optimization.

ICCOPT is my favorite conference so far. It was completely devoted to optimization and the density of interesting talks was overwhelming, very often I could not decide to which parallel session I should go. I was lucky to meet with many brilliant people, whose work inspired some of my papers. I find the diversity of the presented topics quite high, and it is surprising how many different ideas and applications are covered by mathematical optimization.

Organizing was quite fun as well. Next time I may try to organize a couple more sessions to see if it is even more fun at a larger scale.

My 3 sessions were:

  1. Variational Inequalities, Minimax Problems and GANs (Part 1, Monday). Speakers: Simon Lacoste-Julien, Dmitry Kovalev, Gauthier Gidel.

  2. Stochastic Methods for Nonsmooth Optimization (Tuesday). Speakers: Tianyi Lin, Adil Salim and me.

  3. Variational Inequalities, Minimax Problems and GANs (Part 2, Thursday). Speakers: Daoli Zhu, Sarath Pattathil, Georgios Piliouras.

More details about the conference can be found here:

https://iccopt2019.berlin/

The speakers and the titles of their presentations:

  1. Simon Lacoste-Julien “Negative Momentum for Improved Game Dynamics”

  2. Dmitry Kovalev “Revisiting Stochastic Extragradient Method”

  3. Gauthier Gidel “New Optimization Perspectives on Generative Adversarial Networks”

  4. Daoli Zhu “A Variational Approach on Level sets and Linear Convergence of Variable Bregman Proximal Gradient Method for Nonconvex Optimization Problems”

  5. Sarath Pattathil “A unified analysis of optimistic gradient and extra-gradient methods for saddle point problems”

  6. Georgios Piliouras “Online Optimization in Zero - Sum Games and Beyond: A Dynamical Systems Approach”

  7. Tianyi Lin “On Gradient Descent Ascent for Nonconvex-Concave Minimax Optimization”

  8. Adil Salim “Stochastic Proximal Langevin Algorithm”

  9. Konstantin Mishchenko “Variance Reduction for Sums with Smooth and Nonsmooth Components with Linear Convergence”

P.S. Staying in Berlin was as fanstastic as the conference, it is a lovely city and German beer is very helpful for fruitful discussions! :)

Konstantin Mishchenko
Konstantin Mishchenko
Research Scientist

I study optimization and its applications in machine learning.