|
|
TD-Gammon is a backgammon learning program that was developed by the research unit of IBM. In an article at IBM research website Gerald Tesauro, one of IBM research staff, explains that "TD-Gammon is a neural network that trains itself to be an evaluation function for the game of backgammon by playing against itself and learning from the outcome." The TD-Gammon program was developed as a part of a research in the field of reinforcement learning. However, the backgammon program, in Tesauro words "has greatly surpassed all previous computer programs in its ability to Play backgammon". Continue reading to learn about the reasons for TD-Gammon superiority over other backgammon programs. TD-Gammon was not the first learning programmed developed to investigate ideas in artificial intelligence and reinforcement learning. It is also not the first program designed to play backgammon in the level of a human backgammon expert. So, what is the uniqueness of TD-Gammon? According to Tesauro's artice, "TD-Gammon represents a radically different approach toward developing a program capable of sophisticated positional judgment." He explains that TD-Gammon was not programmed to imitate human thinking but to acquire "its own sense of positional judgment by learning from experience in playing against itself." The result, is "an incredibly sophisticated evaluation function which, in at least some cases, appears to surpass the positional judgment of world-class human players." Gerald Tesauro full article on "Temporal Difference Learning and TD-Gammon." GammOnLine Articles on Backgammon Softwares: |