Poker as a Testbed for Machine Knowledge Research.

Uploaded on:
Category: Animals / Pets
Amusement Researchers utilized Chess & other tabletop games as TestBed ... work has concerned different ways to deal with poker diversion tree seek systems, and in addition approaches to ...
Slide 1

By Darse Billings, Dennis Papp, Jonathan Schaeffer, Duane Szafron Poker as a Testbed for Machine Intelligence Research Presented By:- Debraj Manna Gada Kekin Dhiraj Raunak Pillani

Slide 2

CONTENT Introduction Characteristics of Poker Game Texas Hold\'Em Requirements From Players Lokibot Experiment Future Work

Slide 3

INTRODUCTION Game Researchers utilized Chess & other tabletop games as TestBed Poker can be a superior testbed for basic leadership issues

Slide 4

POKER Game of Imperfect learning Risk administration Agent demonstrating Unreliable data Deception Heuristic Search and assessment techniques utilized in Chess not supportive.

Slide 5


Slide 6

TEXAS HOLD Them Pre-Flop – Each player is managed two cards with their face down Community Cards are managed in 3 phases:- Flop – 3 cards are managed face up Turn – 4 th group card is managed face up. Stream – last group card is managed A round of wagering held at every stage Showdown – player having the best 5 cards wins

Slide 7

BETTING STRATEGY FOLD – Withdraw from the amusement CALL – Match the present wager RAISE – Raise the current remarkable wager Only 3 brings are permitted up in a round.

Slide 8

REQUIREMENT Hand Strength – quality of your hand contrasted with adversaries. Hand Potential – Probability of hand enhancing as extra cards show up. Wagering Strategy – Determining ideal wagering methodology Bluffing – Allows you make benefit even on powerless hands

Slide 9

REQUIREMENT (contd.) ‏ Opponent Modeling – Determining likelihood circulation for rivals procedure. Eccentrics – making troublesome for rival to demonstrate your methodology.

Slide 10

Lokibot (later changed to Pokibot) ‏

Slide 11

Pre-flop Evaluation 52 pick 2 = 1326 conceivable blends for two cards Approximate salary rate for every beginning hand utilizing a reenactment of 1,000,000 poker amusements done against nine irregular rivals Highest pay rate: A couple of aces Lowest pay rate: 2 and 7 (of various suits) ‏ One time assessment

Slide 12

Hand Evaluation Hand Strength Assessment of the present quality of the hand Enumeration procedures can give a precise appraisal of the likelihood of right now holding the most grounded hand. Hand Potential changes close by quality

Slide 13

Hand Strength Starting hand is and the lemon is 47 staying obscure cards and {47 pick 2} = 1,081 conceivable hands an adversary may hold. Hand quality is evaluated by basically checking number of conceivable hands that are: superior to our own (any pair, two sets, A-K, or three of a kind: 444 hands) equivalent to our own (9 conceivable residual A-Q blends) more regrettable than our own (628) ‏

Slide 14

Hand Potential Hand quality alone is lacking to survey the nature of a hand Example Hand: Flop: Next card: , Positive/Negative Potential

Slide 15

Hand Potential (contd.) ‏

Slide 16

Hand Potential (contd.) ‏ If T{row,col} alludes to the qualities in the table (B, T, An, and S are Behind, Tied, Ahead, and Sum, resp.) then Ppot and Npot are computed by: Ppot = (T{B,A} + T{B,T}/2 + T{T,A}/2 )/( T{B,S} + T{T,S}/2) ‏ Npot = (T{A,B} + T{A,T}/2 + T{T,B}/2 )/( T{A,S} + T{T,S}/2) ‏ Ppot = 0.208 and Npot = 0.274

Slide 17

Betting Strategy Hand quality and potential are joined into successful hand quality (EHS): EHS = HSn + (1 - HSn ) x Ppot where HSn is the balanced hand quality for n rivals, Ppot is the positive potential. EHS is the likelihood that we are ahead, and in those situations where we are behind there is a Ppot chance that we will pull ahead pot_odds = bets_to_us/( bets_in_pot + bets_to_us ) ‏ Call when Ppot > pot_odds

Slide 18

Experiment Player An is the most developed variant of the project Player E is a fundamental Player B does not have a proper weighting of subcases, utilizing a uniform conveyance for all conceivable rival hands. Player C utilizes an oversimplified pre-flop hand determination strategy, as opposed to the propelled framework which represents player position and number of adversaries. Player D does not have the calculation of hand potential, which is utilized as a part of changing the powerful hand quality and calling with legitimate pot chances.

Slide 19

Experiment (contd.) ‏

Slide 20

Experiment (contd.) ‏ The Bot was additionally keep running against other Poker playing bots and human players over the web. In it\'s present express the bot indicated misfortunes against cutting edge players

Slide 21

Work In Progress It is an anticipated player that responds the same in a given circumstance independent of any chronicled data Opponent displaying: When Lokibot is better ready to deduce likely possessions for the rival, it will be able to do much better choices Betting procedure: feign with high potential hands and sporadically wager a solid hand feebly

Slide 22

Work Done After The Paper Later forms utilized reproduction to find the right move to make, recreating what the activities of alternate players (assessed utilizing the adversary demonstrating) would rely on upon the activity that Lokibot picked. They included particular testing recreation: Opponent displaying comprised of weights for every opening card mix portraying the probabilities of every activity (wager, call, fold) and they quantified rivals by their rate of every activity. The latest work has concerned different ways to deal with poker amusement tree seek strategies, and also approaches to assess perfomance of operators

Slide 23

Contributions Of This Paper Showing that poker can be a testbed of true basic leadership, Identifying the real prerequisites of superior poker, Presenting new specification methods for hand-quality and potential, and Demonstrating a working system that effectively plays "real" poker.

Slide 24

REFERENCE Billings D., Papp D., Schaeffer J. what\'s more, Szafron D. " Poker as a Testbed for Machine Intelligence Research ." In Advances in Artificial Intelligence (Mercer R. what\'s more, Neufeld E. eds.), Springer-Verlag, pp 1-15, 1998.