Toward the understanding of partial-monitoring games thumbnail
Pause
Mute
Subtitles
Playback speed
0.25
0.5
0.75
1
1.25
1.5
1.75
2
Full screen

Toward the understanding of partial-monitoring games

Published on Jul 25, 20114502 Views

Partial monitoring games form a common ground for problems such as learning with expert advice, the multi-armed bandit problem, dynamic pricing, the dark pool problem, label efficient prediction, chan

Related categories

Chapter list

Toward the understanding of partial-monitoring games00:00
Outline01:04
Outline: Partial Monitoring (what and why?) (1)01:30
Prediction Games01:32
A Mathematical Framework03:11
Measure of Performance of the Learner04:06
Outline: Partial Monitoring (what and why?) (2)06:06
What is Missing?06:16
Partial-Monitoring Games11:31
Mathematical Framework12:40
Examples I13:44
Examples II13:53
Example IV: Multi-Armed Bandits16:35
Example V: Dynamic Pricing17:30
Previous Results about Finite Games (1)17:40
Previous Results about Finite Games (2)21:17
Our Contributions22:49
Outline: Results (1)24:18
What Information Can We Collect?24:24
What is the Information Good For?27:22
Outline: Results (2)29:29
The Fundamental Lemma29:31
Proof of the Fundamental Lemma30:10
Proof31:37
The Fundamental Lemma - Illustration31:47
Cell decomposition33:50
Outline: Results (3)34:53
Lower Bound34:56
Lower Bound - Illustration35:37
Outline: Results (4)35:51
Toward the Upper Bound35:51
Constructing Good Estimates37:38
When Should the Learner Stop Using an Action?39:43
Finishing the Upper Bound40:18
Algorithm – BALATON40:47
The Main Result41:40
Summary for Finite, Stochastic Games41:44
Conclusions42:23
For Further Reading43:39