Neural MoveMap Heuristic
Home * Search * Move Ordering * Neural MoveMap Heuristic
Neural MoveMap Heuristic, (NMM heuristic)
a move ordering heuristic introduced by Levente Kocsis at al., applied to the games of Lines of Action [2] and chess [3]. The NMM heuristic utilizes a neural network, trained to estimate the likelihood of a move being the best one in a certain chess position. The best performing NN architecture resulted from an input layer consisting of 6 neurons per square (for each particular piece, -1 Black, 0-None, +1 White) plus one neuron for the side to move, and an butterfly output layer of 1792 neurons corresponding to valid 64x64 from- and to-squares of the moves. Although the network is very large, the move scores can be computed quickly, since one only has to propagate the activation for the pieces actually on the board, and to compute only the scores for the legal moves. An enhanced approach of the NMM heuristic used a weighted combination of the neural network and the history-heuristic scores, tested for middle-game chess positions with Crafty, yielding in encouraging results using Rprop as learning algorithm during the training phase with 4 times (from ECO A30, B84, D85 and E97) 3000 positions [4].
See also
- Butterfly Boards
- Chessmaps Heuristic
- Guard Heuristic
- History Heuristic
- Neural Networks
- Temporal Difference Learning
Publications
- Levente Kocsis, Jos Uiterwijk, Jaap van den Herik (2001). Move Ordering using Neural Networks. IEA/AIE 2001, LNCS 2070
- Levente Kocsis, Jos Uiterwijk, Eric Postma, Jaap van den Herik (2002). The Neural MoveMap Heuristic in Chess. CG 2002, pdf
- Mark Winands, Levente Kocsis, Jos Uiterwijk, Jaap van den Herik (2002). Temporal difference learning and the Neural MoveMap heuristic in the game of Lines of Action. GAME-ON 2002
- Levente Kocsis (2003). Learning Search Decisions. Ph.D thesis, Maastricht University, pdf
Forum Posts
- Neural MoveMap by Niels Abildskov, CCC, May 17, 2021
External Links
- Backpropagation from Wikipedia
- Feedforward neural network from Wikipedia
- Rprop from Wikipedia
- Brecker Brothers - Sponge, Kristianstad/Åhus Jazzfestivalen, July 1980, YouTube Video
References
- ↑ Depiction of a single-layer feedforward artificial neural network. Arrows originating at x2 are omitted for clairty. This network has p inputs and q outputs. Image by Mcstrother, February 17, 2010, CC BY 3.0, Wikimedia Commons, Artificial neural network from Wikipedia
- ↑ Levente Kocsis, Jos Uiterwijk, Jaap van den Herik (2001). Move Ordering using Neural Networks. IEA/AIE 2001, LNCS 2070
- ↑ Levente Kocsis, Jos Uiterwijk, Eric Postma, Jaap van den Herik (2002). The Neural MoveMap Heuristic in Chess. CG 2002, pdf
- ↑ Levente Kocsis, Jos Uiterwijk, Eric Postma, Jaap van den Herik (2002). The Neural MoveMap Heuristic in Chess. CG 2002, pdf