0.25
0.5
0.75
1.25
1.5
1.75
2
Minimax Policies for Combinatorial Prediction Games
Published on Aug 02, 20115878 Views
We address the online linear optimization problem when the actions of the forecaster are represented by binary vectors. Our goal is to understand the magnitude of the minimax regret for the worst po
Related categories
Chapter list
Minimax Policies for Combinatorial Prediction Games00:00
Path planning00:08
Combinatorial prediction game - 101:14
Combinatorial prediction game - 201:29
Combinatorial prediction game - 301:35
Combinatorial prediction game - 401:40
Combinatorial prediction game - 502:05
Combinatorial prediction game - 602:22
Combinatorial prediction game - 702:48
Combinatorial prediction game - 803:10
Notation03:40
Loss assumptions05:46
Key idea07:20
Expanded Exponentially weighted average forecaster (Exp2)08:31
Legendre function11:11
Bregman divergence11:46
CLEB (Combinatorial LEarning with Bregman divergences) - 112:11
CLEB (Combinatorial LEarning with Bregman divergences) - 212:49
CLEB (Combinatorial LEarning with Bregman divergences) - 313:02
CLEB (Combinatorial LEarning with Bregman divergences) - 413:18
CLEB (Combinatorial LEarning with Bregman divergences) - 513:39
CLEB (Combinatorial LEarning with Bregman divergences) - 613:47
General regret bound for CLEB14:35
Di erent instances of CLEB: LinExp (Entropy Function) 15:03
Di erent instances of CLEB: LinINF (Exchangeable Hessian)16:08
Di erent instances of CLEB: Follow the regularized leader16:47
Minimax regret for combinatorial prediction games17:15