CCIMI Seminar

With Sébastien Bubeck (Microsoft Research Redmond)

CCIMI Seminar: Kernel-based Methods for Bandit Convex Optimization

A lot of progress has been made in recent years on extending classical multi-armed bandit strategies to very large set of actions. A particularly challenging setting is the one where the action set is continuous and the underlying cost function is convex, this is the so-called bandit convex optimization (BCO) problem. I will tell the story of BCO and explain some of the new ideas that we recently developed to solve it. I will focus on three new ideas from our recent work with Yin Tat Lee and Ronen Eldan: (i) a new connection between kernel methods and the popular multiplicative weights strategy; (ii) a new connection between kernel methods and one of Erdos’ favorite mathematical object, the Bernoulli convolution, and (iii) a new adaptive (and increasing!) learning rate for multiplicative weights. These ideas could be of broader interest in learning/algorithm’s design.

This talk is part of the CCIMI Seminar Series

  • Speaker: Sébastien Bubeck (Microsoft Research Redmond)
  • Wednesday 04 October 2017, 16:0017:00
  • Venue: MR12.
  • Series: Statistics; organiser: Quentin Berthet.

Add to your calendar or Include in your list