Programming backgammon using self-teaching neural nets

Artificial Intelligence

Volumn 134, Issue 1-2, 2002, Pages 181-199

  Tesauro, Gerald ^a  

^a IBM T J WATSON RESEARCH CENTER   (United States)

Author keywords

Backgammon; Doubling strategy; Games; Neural networks; Reinforcement learning; Rollouts; Temporal difference learning

Indexed keywords

ALGORITHMS; ARTIFICIAL INTELLIGENCE; COMPUTER PROGRAM LISTINGS; LEARNING SYSTEMS; NEURAL NETWORKS;

SELF-TEACHING NEURAL NETS;

GAME THEORY;

EID: 0036147771     PISSN: 00043702     EISSN: None     Source Type: Journal    
DOI: 10.1016/S0004-3702(01)00110-2     Document Type: Article

References (31)

1. A knowledge-based approach of Connect-Four
2. KnightCap: A chess program that learns by combining TD(λ) with minimax search
3. Computer backgammon
4. Efficient approximation of backgammon race equities
5. Improving elevator performance using reinforcement learning
6. Multilayer feedforward networks are universal approximators
7. Optimal doubling in backgammon
8. Optimizing production manufacturing using reinforcement learning
9. Call admission control and routing in integrated service networks using neuro-dynamic programming
10. Reinforcement learning for trading systems and portfolios
11. Co-evolution in the successful learning of backgammon strategy
12. Carbon versus silicon: Matching wits with TD-Gammon
13. Learning internal representation by error propagation
14. Some studies in machine learning using the game of checkers
15. The games computers (and people) play
16. Value function based production scheduling
17. Reinforcement learning for dynamic channel allocation in cellular telephone systems
18. Learning to predict by the methods of temporal differences
19. Neurogammon wins computer olympiad
20. Practical issues in temporal difference learning
21. Temporal difference learning and TD-Gammon
22. On-line policy improvement using Monte-Carlo search
23. On optimal doubling in backgammon
24. High-performance job-shop scheduling with a time-delay TD(λ) network