A cellular simultaneous recurrent network (CSRN) [1-11] is a neural network architecture that uses conventional simultaneous recurrent networks (SRNs), or cells in a cellular structure. The cellular structure adds complexity, so the training of CSRNs is far more challenging than that of conventional SRNs. Computer Go serves as an excellent test bed for CSRNs because of its clear-cut objective. For the training data, we developed an accurate theoretical foundation and game tree for the 2x2 game board. The conventional CSRN architecture suffers from the multi-valued function problem; our modified CSRN architecture overcomes the problem by employing ternary coding of the Go board's representation and a normalized input dimension reduction. We demonstrate a 2x2 game tree trained with the proposed CSRN architecture and the proposed cellular particle swarm optimization.
T. Kim and D. C. Wunsch, "Modified Cellular Simultaneous Recurrent Networks with Cellular Particle Swarm Optimization," Proceedings of the International Joint Conference on Neural Networks, 2012, Institute of Electrical and Electronics Engineers (IEEE), Jan 2012.
The definitive version is available at http://dx.doi.org/10.1109/IJCNN.2012.6252845
2012 Annual International Joint Conference on Neural Networks, IJCNN '12, Part of the 2012 IEEE World Congress on Computational Intelligence, WCCI 2012 (2012: Jun. 10-15, Brisbane, Australia)
Electrical and Computer Engineering
International Standard Book Number (ISBN)
Article - Conference proceedings
© 2012 Institute of Electrical and Electronics Engineers (IEEE), All rights reserved.