0.25
0.5
0.75
1.25
1.5
1.75
2
Opportunistic Strategies for Generalized No-Regret Problems
Published on Aug 09, 20133661 Views
No-regret algorithms has played a key role in on-line learning and prediction problems. In this paper, we focus on a generalized no-regret problem with vector-valued rewards, defined in terms of a des
Related categories
Chapter list
Opportunistic Strategies for Generalized No-Regret Problems00:00
Outline - 100:01
Outline - 200:18
Generalized No-Regret Problems00:19
Generalized No-Regret Algorithms01:17
Example: Constrained Regret Minimization03:10
Outline - 305:50
Approachability Problem05:57
Approachability Conditions08:14
Outline - 410:37
Calibrated Forecasts10:47
The Algorithm12:31
Outline - 514:31
Desiderata for Opportunistic No-Regret Algorithm14:33
Restricted Opponent’s Play - 115:18
Restricted Opponent’s Play - 216:10
Examples and Properties16:40
Restricted Target Set - 116:44
Restricted Target Set - 217:34
Opportunistic Strategies17:47
Main Results18:22
Proof Idea19:51
Computational Issues and Convergence Rates19:57