counterfactual regret minimization and domination in
play

Counterfactual Regret Minimization and Domination in Extensive-Form - PowerPoint PPT Presentation

Dec 23, 2023 •115 likes •479 views

Counterfactual Regret Minimization and Domination in Extensive-Form Games Richard Gibson University of Alberta Edmonton, Alberta, Canada Overview Counterfactual Regret Minimization (CFR) Overview Counterfactual Regret Minimization (CFR)

  1. Counterfactual Regret Minimization and Domination in Extensive-Form Games Richard Gibson University of Alberta Edmonton, Alberta, Canada

  2. Overview Counterfactual Regret Minimization (CFR)

  3. Overview Counterfactual Regret Minimization (CFR) Provably solves for Nash equilibrium 2-Player Zero-Sum Extensive-Form Games

  4. Overview Counterfactual Regret Minimization (CFR) Provably solves for Seems to work well... Nash equilibrium 2-Player Zero-Sum Extensive-Form Extensive-Form Games, any Games Number of players

  5. Overview Counterfactual Regret Minimization (CFR) Provably solves for Seems to work well... Nash equilibrium 2-Player Zero-Sum Extensive-Form Extensive-Form Games, any Games Number of players Question : Why do CFR strategies work well in extensive-form games outside of the 2-player zero-sum case?

  6. Extensive-Form Games C QJ QK 0.5 0.5 1 1 c b c b 2 2 2 2 c b f c c b f c (1,-1) 1 (1,-1) (2,-2) (-1,1) -1 1 (1,-1) (-2,2) f c f c (-1,1) (2,-2) (-1,1) (-2,2)

  7. Extensive-Form Games C QJ QK 0.5 0.5 1 1 c b c b 2 2 2 2 c b f c c b f c (1,-1) 1 (1,-1) (2,-2) (-1,1) -1 1 (1,-1) (-2,2) f c f c (-1,1) (2,-2) (-1,1) (-2,2) Information sets group states that are indistinguishable to the player.

  8. Extensive-Form Games C QJ QK 0.5 0.5 1 1 0.4 0.6 0.4 0.6 2 2 2 2 0.7 0.3 0 1 0.9 0.1 1 0 (1,-1) 1 (1,-1) (2,-2) (-1,1) 1 (1,-1) (-2,2) 0.2 0.8 0.2 0.8 (-1,1) (2,-2) (-1,1) (-2,2) A strategy profile σ = ( σ 1 , σ 2 ) assigns a probability distribution over actions at each information set. Example: Probability player 1 checks is σ 1 ( Q?, c ) = 0.4 .

  9. Extensive-Form Games C QJ QK 0.5 0.5 1 1 0.4 0.6 0.4 0.6 2 2 2 2 0.7 0.3 0 1 0.9 0.1 1 0 (1,-1) 1 (1,-1) (2,-2) (-1,1) 1 (1,-1) (-2,2) 0.2 0.8 0.2 0.8 (-1,1) (2,-2) (-1,1) (-2,2) u i ( σ ) is the expected utility for player i , assuming players play according to σ .

  10. Counterfactual Regret Minimization (CFR) [Zinkevich et al ., NIPS 2007] ● CFR is an iterative algorithm that generates strategy profiles ( σ 1 , σ 2 , ... , σ T ) over many iterations T . ● Final output of CFR: σ AVG = Average( σ 1 , σ 2 , ... , σ T ). ● For 2-player zero-sum games, σ AVG is an ϵ -Nash equilbrium, with ϵ → 0 as T → ∞: AVG , σ 2 AVG ) ≥ max u 1 ( σ 1 * , σ 2 AVG ) - ϵ u 1 ( σ 1 * σ 1 AVG , σ 2 AVG ) ≥ max u 2 ( σ 1 AVG , σ 2 * ) - ϵ u 2 ( σ 1 σ 2 *

  11. Counterfactual Regret Minimization (CFR) ● Outside of 2-player zero-sum games, σ AVG is not necessarily an approximate Nash equilibrium [Abou Risk and Szafron, AAMAS 2010] . – A player may gain by deviating from σ AVG . ● In these games, a Nash equilibrium might not be the most appropriate solution concept anyways. ● On the other hand, σ AVG performs very well in practice...

  12. Annual Computer Poker Competition 3-Player Limit Hold'em - 2009 3-Player Limit Hold'em - 2011 Agent Instant Run-off: Round 0 Agent Instant Run-off: Round 0 Hyperborean-Eqm 319 ± 2 Sartre3p 243 ± 20 Hyperborean-BR 299 ± 2 Hyperborean-3p-limit-iro 204 ± 20 akuma 151 ± 2 LittleRock 113 ± 19 dpp 171 ± 2 AAIMontybot 96 ± 44 CMURingLimit -37 ± 2 dcubot3plr 77 ± 19 dcu3pl -63 ± 2 OwnBot -4 ± 30 Bluechip -548 ± 2 Bnold3 -91 ± 22 Entropy -108 ± 36 3-Player Limit Hold'em - 2010 player.zeta.3p -530 ± 33 Agent Instant Run-off: Round 0 Hyperborean.iro 144 ± 32 dcu3pl.tbr 98 ± 30 LittleRock 65 ± 35 Arnold3 -135 ± 39 Bender -172 ± 16

  13. Counterfactual Regret Minimization (CFR) ● In games with more than 2-players, σ AVG is a “good” strategy. Why? ● What properties make a strategy good in games with more than 2-players? ● We know what a bad strategy is...

  14. Domination C QJ QK 0.5 0.5 1 1 c b c b 2 2 2 2 c b f c c b f c (1,-1) 1 (1,-1) (2,-2) (-1,1) -1 1 (1,-1) (-2,2) f c f c (-1,1) (2,-2) (-1,1) (-2,2)

  15. Domination C QJ QK 0.5 0.5 1 1 c b c b 2 2 2 2 c b 0 1 c b f c (1,-1) 1 (1,-1) (2,-2) (-1,1) -1 1 (1,-1) (-2,2) f c f c (-1,1) (2,-2) (-1,1) (-2,2) J, c that always calls with the Jack when faced Consider any player 2 strategy σ 2 with a bet.

  16. Domination C QK QJ 0.5 0.5 1 1 c b c b 2 2 2 2 c b 0 1 c b f c (1,-1) 1 (1,-1) (2, -2 ) (-1,1) -1 1 (1,-1) (-2,2) f c f c (-1,1) (2,-2) (-1,1) (-2,2) J, c ) = ... + 0.5 σ 1 ( Q?, b )1 ( -2 ) + ... u 2 ( σ 1 , σ 2

  17. Domination C QK QJ 0.5 0.5 1 1 c b c b 2 2 2 2 c b 1 0 c b f c (1,-1) 1 (1,-1) (2,-2) (-1,1) -1 1 (1,-1) (-2,2) f c f c (-1,1) (2,-2) (-1,1) (-2,2) J, c ) = ... + 0.5 σ 1 ( Q?, b )1 ( -2 ) + ... u 2 ( σ 1 , σ 2 J, f . Now consider the same player 2 strategy, except always folds the J. Call it σ 2

  18. Domination C QK QJ 0.5 0.5 1 1 c b c b 2 2 2 2 c b 1 0 c b f c (1,-1) 1 (1, -1 ) (2,-2) (-1,1) -1 1 (1,-1) (-2,2) f c f c (-1,1) (2,-2) (-1,1) (-2,2) J, c ) = ... + 0.5 σ 1 ( Q?, b )1 ( -2 ) + ... u 2 ( σ 1 , σ 2 J, f ) = '' + 0.5 σ 1 ( Q?, b )1 ( -1 ) + '' u 2 ( σ 1 , σ 2

  19. Domination C QK QJ 0.5 0.5 1 1 c b c b 2 2 2 2 c b 1 0 c b f c (1,-1) 1 (1, -1 ) (2,-2) (-1,1) -1 1 (1,-1) (-2,2) f c f c (-1,1) (2,-2) (-1,1) (-2,2) J, c ) = ... + 0.5 σ 1 ( Q?, b )1 ( -2 ) + ... u 2 ( σ 1 , σ 2 J, f ) = '' + 0.5 σ 1 ( Q?, b )1 ( -1 ) + '' u 2 ( σ 1 , σ 2 J, c ) ≤ u 2 ( σ 1 , σ 2 J, f ) for all σ 1 . u 2 ( σ 1 , σ 2 J, c ) < u 2 ( σ 1 , σ 2 J, f ) if σ 1 ( Q?, b ) > 0 u 2 ( σ 1 , σ 2

  20. Domination C QJ QK 0.5 0.5 1 1 c b c b 2 2 2 2 c b f c c b f c (1,-1) 1 (1,-1) (2,-2) (-1,1) -1 1 (1,-1) (-2,2) f c f c (-1,1) (2,-2) (-1,1) (-2,2) σ 2 is a dominated strategy if there exists σ 2 ' such that u 2 ( σ 1 , σ 2 , σ 3 , ... ) ≤ u 2 ( σ 1 , σ 2 ' , σ 3 , ... ) for all σ 1 , σ 3 , ... u 2 ( σ 1 , σ 2 , σ 3 , ... ) < u 2 ( σ 1 , σ 2 ' , σ 3 , ... ) for some σ 1 , σ 3 , ...

  21. Domination C QJ QK 0.5 0.5 1 1 c b c b 2 2 2 2 c b f c c b f c (1,-1) 1 (1,-1) (2,-2) (-1,1) -1 1 (1,-1) (-2,2) f c f c (-1,1) (2,-2) (-1,1) (-2,2) J, c is dominated by σ 2 J, f σ 2 K, f is dominated by σ 2 K, c σ 2

  22. Domination C QJ QK 0.5 0.5 1 1 c b c b 2 2 2 2 c b f c c b f c (1,-1) 1 (1,-1) (2,-2) (-1,1) -1 1 (1,-1) (-2,2) f c f c (-1,1) (2,-2) (-1,1) (-2,2) Define a dominated action to be an action such that any strategy that always plays that action is dominated (assuming that player plays to reach that action).

  23. Domination C QJ QK 0.5 0.5 1 1 c b c b 2 2 2 2 c b f c b c (1,-1) 1 (1,-1) (-1,1) -1 1 (-2,2) f c f c (-1,1) (2,-2) (-1,1) (-2,2)

  24. Domination C QJ QK 0.5 0.5 1 1 0 1 0 1 2 2 2 2 c b f c b c (1,-1) 1 (1,-1) (-1,1) -1 1 (-2,2) f c f c (-1,1) (2,-2) (-1,1) (-2,2) b that always bets. Consider the player 1 strategy σ 1

  25. Domination C QJ QK 0.5 0.5 1 1 0 1 0 1 2 2 2 2 c b 1 c b 1 (1,-1) 1 ( 1 ,-1) (-1,1) -1 1 ( -2 ,2) f c f c (-1,1) (2,-2) (-1,1) (-2,2) b , σ 2 ) = 0.5 ( 1 )( 1 )( 1 ) + 0.5 ( 1 )( 1 )( -2 ) = -0.5 u 1 ( σ 1

  26. Domination C QJ QK 0.5 0.5 1 1 1 0 1 0 2 2 2 2 c b f c b c (1,-1) 1 (1,-1) (-1,1) -1 1 (-2,2) 0 1 0 1 (-1,1) (2,-2) (-1,1) (-2,2) b , σ 2 ) = 0.5 ( 1 )( 1 )( 1 ) + 0.5 ( 1 )( 1 )( -2 ) = -0.5 u 1 ( σ 1 cc that checks then calls. Now consider the player 1 strategy σ 1

  27. Domination C QJ QK 0.5 0.5 1 1 1 0 1 0 2 2 2 2 1 0 f 0 1 c ( 1 ,-1) 1 (1,-1) (-1,1) -1 1 (-2,2) 0 1 0 1 (-1,1) (2,-2) (-1,1) ( -2 ,2) b , σ 2 ) = 0.5 ( 1 )( 1 )( 1 ) + 0.5 ( 1 )( 1 )( -2 ) = -0.5 u 1 ( σ 1 cc , σ 2 J c ,K b ) = 0.5 ( 1 )( 1 )( 1 ) + 0.5 ( 1 )( 1 )( 1 )( -2 ) = -0.5 u 1 ( σ 1

  28. Domination C QJ QK 0.5 0.5 1 1 1 0 1 0 2 2 2 2 0 1 f 1 0 c (1,-1) 1 (1,-1) ( -1 ,1) -1 1 (-2,2) 0 1 0 1 (-1,1) ( 2 ,-2) (-1,1) (-2,2) b , σ 2 ) = 0.5 ( 1 )( 1 )( 1 ) + 0.5 ( 1 )( 1 )( -2 ) = -0.5 u 1 ( σ 1 cc , σ 2 J c ,K b ) = 0.5 ( 1 )( 1 )( 1 ) + 0.5 ( 1 )( 1 )( 1 )( -2 ) = -0.5 u 1 ( σ 1 cc , σ 2 J b ,K c ) = 0.5 ( 1 )( 1 )( 1 )( 2 ) + 0.5 ( 1 )( 1 )( -1 ) = +0.5 u 1 ( σ 1

  29. Domination C QJ QK 0.5 0.5 1 1 c b c b 2 2 2 2 c b f c b c (1,-1) 1 (1,-1) (-1,1) -1 1 (-2,2) f c f c (-1,1) (2,-2) (-1,1) (-2,2) σ 1 is an iteratively dominated strategy if there exists σ 1 ' such that u 1 ( σ 1 , σ 2 , σ 3 , ... ) ≤ u 1 ( σ 1 ' , σ 2 , σ 3 , ... ) for all non-iteratively dominated σ 2 , σ 3 , ... u 1 ( σ 1 , σ 2 , σ 3 , ... ) < u 1 ( σ 1 ' , σ 2 , σ 3 , ... ) for some non-iteratively dominated σ 2 , σ 3 , ...

  30. Domination C QJ QK 0.5 0.5 1 1 c b c b 2 2 2 2 c b f c b c (1,-1) 1 (1,-1) (-1,1) -1 1 (-2,2) f c f c (-1,1) (2,-2) (-1,1) (-2,2) b is iteratively dominated by σ 1 cc σ 1

  31. Domination C QJ QK 0.5 0.5 1 1 c b c b 2 2 2 2 c b f c b c (1,-1) 1 (1,-1) (-1,1) -1 1 (-2,2) f c f c (-1,1) (2,-2) (-1,1) (-2,2) Define an iteratively dominated action to be an action such that any strategy that always plays that action is iteratively dominated (assuming that player plays to reach that action).

Recommend

Counterfactual Regret Minimization Gabriele Farina 1 Christian Kroer 2 Noam Brown 1 Tuomas Sandholm

Counterfactual Regret Minimization Gabriele Farina 1 Christian Kroer 2 Noam Brown 1 Tuomas Sandholm

Stable-Predictive Optimistic Counterfactual Regret Minimization Gabriele Farina 1 Christian Kroer 2 Noam Brown 1 Tuomas Sandholm 1,3 1 Computer Science Department, Carnegie Mellon University 2 IEOR Department, Columbia University 3 Strategic

283 views • 5 slides
Deep Counterfactual Regret Min inimization Noam Brown* 12 , Adam Lerer* 1 , Sam Gross 1 , Tuomas

Deep Counterfactual Regret Min inimization Noam Brown* 12 , Adam Lerer* 1 , Sam Gross 1 , Tuomas

Deep Counterfactual Regret Min inimization Noam Brown* 12 , Adam Lerer* 1 , Sam Gross 1 , Tuomas Sandholm 23 *Equal Contribution 1 Facebook AI Research 2 Carnegie Mellon University 3 Strategic Machine Inc., Strategy Robot Inc., and Optimized

487 views • 11 slides
Cautious R Regret M Minimization: Online O Optimization w with L Long-Term B Budg udget Co

Cautious R Regret M Minimization: Online O Optimization w with L Long-Term B Budg udget Co

Cautious R Regret M Minimization: Online O Optimization w with L Long-Term B Budg udget Co Constraints Nikolaos Liakopoulos 1,2,3 , Apostolos Destounis 1 , Georgios S. Paschos 1 , Thrasyvoulos Spyropoulos 2 , Panayotis Mertikopoulos 4 1

315 views • 13 slides
Optimistic Regret Minimization for Extensive-Form Games via Dilated Distance-Generating Functions

Optimistic Regret Minimization for Extensive-Form Games via Dilated Distance-Generating Functions

Optimistic Regret Minimization for Extensive-Form Games via Dilated Distance-Generating Functions Gabriele Farina 1 Christian Kroer 2 Tuomas Sandholm 1,3,4,5 1 Computer Science Department, Carnegie Mellon University 2 IEOR Department, Columbia

512 views • 32 slides
U G A V ! Michael Johanson, Nolan Bard, Marc Lanctot, &quot; # ! K Q $ A J Richard

U G A V ! Michael Johanson, Nolan Bard, Marc Lanctot, &quot; # ! K Q $ A J Richard

Efficient Nash Equilibrium Approximation through Monte Carlo Counterfactual Regret Minimization AAMAS 2012 - June 6, 2012 Q J # $ K 1 0 P C R &quot; ! U G A V ! Michael Johanson, Nolan Bard, Marc Lanctot, &quot; # ! K Q $

539 views • 24 slides
Bayesian Counterfactual Risk Minimization Ben London (blondon@) Amazon Music Ted Sandler (sandler@)

Bayesian Counterfactual Risk Minimization Ben London (blondon@) Amazon Music Ted Sandler (sandler@)

Bayesian Counterfactual Risk Minimization Ben London (blondon@) Amazon Music Ted Sandler (sandler@) Amazon Music International Conference on Machine Learning Long Beach, CA, June 11, 2019 Learning from Logged Data Pull log data e.g., user i

593 views • 10 slides
Regret Minimization for Online Buffering Problems Using the Weighted Majority Algorithm Sascha

Regret Minimization for Online Buffering Problems Using the Weighted Majority Algorithm Sascha

Regret Minimization for Online Buffering Problems Using the Weighted Majority Algorithm Sascha Geulen, Berthold V ocking, Melanie Winkler Department of Computer Science, RWTH Aachen University June 27, 2010 Melanie Winkler (RWTH Aachen

240 views • 11 slides
Two domination parameters in graphs Guangjun Xu Department of Mathematics and Statistics The

Two domination parameters in graphs Guangjun Xu Department of Mathematics and Statistics The

Two domination parameters in graphs Guangjun Xu Department of Mathematics and Statistics The University of Melbourne March 17, 2009 Joint work with Liying Kang, Erfang Shan and Min Zhao Domination in graphs Power domination Rainbow

958 views • 77 slides
randregret : A command for fitting random regret minimization models using Stata UK Stata

randregret : A command for fitting random regret minimization models using Stata UK Stata

3 4 3 . 5 2 . 5 2 0 . 5 2 . 5 2 1 . 5 1 0 . 5 = 1 1 . 5 1 = 2 4 0 3 . 5 = 0 . 5 = 0 . 05 = 15 ( x jmn x imn ) r 0 . 5 1 1 . 5 2 2 . 5 3 randregret : A command for fitting random regret

1.87k views • 132 slides
On traffic domination in communication networks Walid Ben-Ameur 1 Pablo Pavon 2 oro 3 , 4 Micha

On traffic domination in communication networks Walid Ben-Ameur 1 Pablo Pavon 2 oro 3 , 4 Micha

introduction notation total domination ordinary domination conclusions On traffic domination in communication networks Walid Ben-Ameur 1 Pablo Pavon 2 oro 3 , 4 Micha Pi 1 TELECOM SudParis, France 2 Technical University of Cartagena,

431 views • 17 slides
Domination in generalized Domination in generalized Petersen graphs Petersen graphs Advisor:

Domination in generalized Domination in generalized Petersen graphs Petersen graphs Advisor:

Domination in generalized Domination in generalized Petersen graphs Petersen graphs Advisor: Sheng-Chyang Liaw Student: Yuan-Shin Li Introduction A vertex subset S of a graph G is a dominating set g p g if each vertex in V ( G ) S is

853 views • 50 slides
Minimization Using Descent Information we will consider the minimization of unconstrained

Minimization Using Descent Information we will consider the minimization of unconstrained

Minimization Using Descent Information we will consider the minimization of unconstrained functions of several variables where we now assume we have some derivative information such as the gradient vector or the Hessian matrix. Recall

131 views • 11 slides
Formulas for various domination numbers of products of paths and cycles Janez c 1 Zerovnik 1 ,

Formulas for various domination numbers of products of paths and cycles Janez c 1 Zerovnik 1 ,

Polygraphs Domination and its variations Path algebra The algorithm Results and remarks Formulas for various domination numbers of products of paths and cycles Janez c 1 Zerovnik 1 , 2 Polona Pavli 1 IMFM, Ljubljana, Slovenia 2 FME,

959 views • 59 slides
Counterfactual Donkeys and the Modal Horizon Andreas Walker and Maribel Romero Counterfactual

Counterfactual Donkeys and the Modal Horizon Andreas Walker and Maribel Romero Counterfactual

Counterfactual Donkeys and the Modal Horizon Andreas Walker and Maribel Romero Counterfactual Donkeys and the Modal Horizon Andreas Walker and Maribel Romero 1 Introduction Counterfactual Donkey Sentences 2 Readings High and low readings 3

593 views • 30 slides
Realizations of the Game Domination Number Bo stjan Bre sar, Sandi Klav zar, Ga sper

Realizations of the Game Domination Number Bo stjan Bre sar, Sandi Klav zar, Ga sper

Realizations of the Game Domination Number Bo stjan Bre sar, Sandi Klav zar, Ga sper Ko smrlj, Doug Rall Faculty of Mathematics and Physics University of Ljubljana, Slovenia 20th September 2012 Domination Game game on a finite

679 views • 28 slides
Equality in the Domination Chain in Triangulations Stephen Finbow Joint work with C. M. van

Equality in the Domination Chain in Triangulations Stephen Finbow Joint work with C. M. van

Definitions and Introduction Well-covered Graphs Equality in the Domination Chain Triangulation Equality in the Domination Chain in Triangulations Stephen Finbow Joint work with C. M. van Bommel St. Francis Xavier University June 12, 2013

699 views • 51 slides
Partial Functions and Domination C T Chong National University of Singapore

Partial Functions and Domination C T Chong National University of Singapore

Partial Functions and Domination C T Chong National University of Singapore chongct@math.nus.edu.sg CTFM, Tokyo 711 September 2015 Domination for Partial Functions Definition Let f , g be partial functions. Then g dominates f if

664 views • 28 slides
Counterfactual Policy Evaluation in Reproducing Kernel Hilbert Spaces Krikamol Muandet Max

Counterfactual Policy Evaluation in Reproducing Kernel Hilbert Spaces Krikamol Muandet Max

Counterfactual Policy Evaluation in Reproducing Kernel Hilbert Spaces Krikamol Muandet Max Planck Institute for Intelligent Systems Tbingen, Germany Jeju, Korea February 22, 2019 Krikamol Muandet Counterfactual Learning in RKHS Jeju,

1.14k views • 61 slides
Domination in circle graphs Nicolas Bousquet Daniel Gon calves George B. Mertzios Christophe

Domination in circle graphs Nicolas Bousquet Daniel Gon calves George B. Mertzios Christophe

Circles graphs Dominating set Some positive results Open Problems Domination in circle graphs Nicolas Bousquet Daniel Gon calves George B. Mertzios Christophe Paul Ignasi Sau St ephan Thomass e Agape 2012 Domination in circle

536 views • 37 slides
Minimization Satoru Iwata (University of Tokyo) Submodular Function Minimization ( )

Minimization Satoru Iwata (University of Tokyo) Submodular Function Minimization ( )

Submodular Function Minimization Satoru Iwata (University of Tokyo) Submodular Function Minimization ( ) 0 f Assumption: X Minimization Evaluation Algorithm Oracle ( X ) f Minimizer min{ ( ) | } ? f Y Y V

177 views • 16 slides
One-Dimensional Minimization Lectures for PHD course on Numerical optimization Enrico Bertolazzi

One-Dimensional Minimization Lectures for PHD course on Numerical optimization Enrico Bertolazzi

One-Dimensional Minimization Lectures for PHD course on Numerical optimization Enrico Bertolazzi DIMS Universit a di Trento November 21 December 14, 2011 One-Dimensional Minimization 1 / 33 Outline Golden Section minimization 1

654 views • 33 slides
Counterfactual-based mediation analysis Workshop 2 Rhian Daniel London School of Hygiene and

Counterfactual-based mediation analysis Workshop 2 Rhian Daniel London School of Hygiene and

Counterfactual-based mediation analysis Workshop 2 Rhian Daniel London School of Hygiene and Tropical Medicine CIMPOD 28th February, 2017 Rhian Daniel/Counterfactual-based mediation analysisWorkshop 2 1/55 Setting the scene Case study

1.85k views • 171 slides
Counterfactual-based mediation analysis Workshop 1 Rhian Daniel London School of Hygiene and

Counterfactual-based mediation analysis Workshop 1 Rhian Daniel London School of Hygiene and

Counterfactual-based mediation analysis Workshop 1 Rhian Daniel London School of Hygiene and Tropical Medicine CIMPOD 27th February, 2017 Rhian Daniel/Counterfactual-based mediation analysisWorkshop 1 1/51 Setting the scene Case study

1.78k views • 131 slides
No-Regret Learning in Convex Games Geoff Gordon, Amy Greenwald, Casey Marks, and Martin Zinkevich

No-Regret Learning in Convex Games Geoff Gordon, Amy Greenwald, Casey Marks, and Martin Zinkevich

No-Regret Learning in Convex Games Geoff Gordon, Amy Greenwald, Casey Marks, and Martin Zinkevich No-Regret Learning in Convex Games p. 1 Introduction The connection between regret and equilibria is well understood in matrix games. Most

554 views • 42 slides

More recommend