Five Miracles of Mirror Descent, Lecture 2/9

Поділитися
Вставка
  • Опубліковано 10 чер 2024
  • Lectures on ``some geometric aspects of randomized online decision making" by Sebastien Bubeck for the summer school HDPA-2019 (High dimensional probability and algorithms) hdpa2019.sciencesconf.org/
    Lecture 2:
    - Multiplicative update algorithm and its regret analysis
    - Continuous time mirror descent and its potential analysis
    Lecture notes can be found here: hdpa2019.sciencesconf.org/dat...

КОМЕНТАРІ • 8

  • @avi111986
    @avi111986 4 роки тому +2

    I'm not sure about Approach 1 suggested around 4:00 . How does he perform GD on the function < l_t, p >, when l_t is not known in advance? If l_t is known to the player at the beginning of time t why not just choose some expert with 0 loss? I'm probably missing something in the problem description. Can someone please help me out here?

    • @SebastienBubeck
      @SebastienBubeck  4 роки тому +4

      The suggestion is to get p_{t+1} from p_t by a step of gradient descent on the function < l_t, p >. In particular, this operation can be performed at the beginning of round t+1 (when you need p_{t+1}), and thus at a time when l_t is known to the player.

  • @scose
    @scose 3 роки тому +1

    Is there a written reference for this Riemannian interpretation of mirror descent? It seems different from the interpretation in your work "Convex Optimization: Algorithms and Complexity", which doesn't mention manifolds.

    • @SebastienBubeck
      @SebastienBubeck  3 роки тому +2

      Unfortunately I have not written it yet, but I have plans to do it at some point in the future... For the moment you can take a look at this paper arxiv.org/abs/2004.01025 , although they have a different interpretation than mine.

    • @scose
      @scose 3 роки тому +1

      @@SebastienBubeck Thank you!

  • @honey-py9pj
    @honey-py9pj Рік тому

    In 9:49 it is stated that opt is a vector that has everywhere 0 expect from one coordinate , let's say the i-th. Why exactly?To find this we take the gradient of cumulative losses for a fixed probability distribution p , right? And then?

  • @christianholtz5182
    @christianholtz5182 Рік тому

    Hi, me again. I start a company called 5k education. Book learning, watch lectures on tv with 3 friends, testing. 2 degrees for 5k (since most fail, see executive function).

  • @cpthddk
    @cpthddk 3 роки тому

    Cameraman kind of sucks on this one... I feel like they didn't get that the content was important, not seb's hot bod