N. Richards, D. Moriarty, and R. Miikkulainen, "Evolving neural networks to play go," Applied Intelligence, vol. 8, pp. 85--96, 1998.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....to evolve [20, 15] 2.5 Simultaneous Evolution One of the most interesting areas of evolutionary neural networks is the combination of several schemes which simultaneously evolve di erent aspects of the networks. One of the most important is the combination of architecture and weight evolution [11, 7, 12, 21, 22, 18, 23, 24, 25]. The advantage of combining these two basic elements of a neural network is that a completely functioning network can be evolved without any human interaction. Clearly, it might be advantageous to simultaneously evolve more neural network features, thus leading to more ecient and accurate ....

N. Richards, D.E. Moriarty, and R. Miikkulainen. Evolving neural networks to play go. Applied Intelligence, 8:85-96, 1997.

....is not likely that a human will play thousands of games against a computer program. Go servers on the Internet might however make it possible to nd human opponents for a smaller number of games. 9 2.7.3 The Opponent The degree of determinism in the opponent a very important factor to consider. [11] Since the program will learn to evaluate positions based on the result of playing out the game against a particular opponent, it is important that this opponent is as good and as varied as possible. If an opponent has a weakness, then the algorithm is likely to learn to exploit that weakness ....

Norman Richards, David Moriarty, and Risto Miikkulainen. Evolving Neural Networks to Play Go, Applied Intelligence, 8:85-96. 1998.

....FACES OF GO) has seen the addition of many other types of reasoning and specialist modules. Non symbolic techniques have been used to learn=evolve controllers and rules based upon patterns of stones for use during play. The techniques applied include genetic programming [10] genetic algorithms [32, 22, 12], and neural networks [13] These approaches have so far been less successful than the hybrid programs but have the advantage that Go knowledge does not need to be added by hand. Cazenave s GOGOL [8] applies learning techniques to good e#ect. An o# line program uses introspection to prove ....

N. Richards, D. Moriarty, R. Miikkulainen, Evolving neural networks to play go, Tech. Rep., The University of Texas at Austin, 1997.

....trivial, its explosive complexity of play has prevented it from succumbing to the sort of brute force search that has proven effective in chess. Probably because go relies greatly on pattern and shape, approaches using neural networks have shown success in the past, especially on smaller boards [4, 5, 10]. This paper first seeks to see how well the ESP neuro evolution algorithm works at developing computer go opponents, because this particular technique has not been tried previously. Then, it explores ways of improving performance by forcing parts of the network to specialize on different areas of ....

....on pattern recognition and intuition about a situation, which makes it hard to codify their expertise. Neural networks are quite effective at pattern recognition and detecting vague properties such as shape and form [2] so some work has been done in recent years concerning their application to go [4, 5, 10]. 3. Neuro evolution Neural networks make use of networks of artificial neurons, which are processing units that compute some simple function of their input values, producing one or more output values. Typical neurons compute a weighted sum of their numerical inputs, then apply a threshold ....

[Article contains additional citation context not shown here]

Norman Richards, David E. Moriarty, and Risto Miikkulainen. Evolving Neural networks to play go. Applied Intelligence, 1998.

....of Go) has seen the addition of many other types of reasoning and specialist modules. ffl Non symbolic techniques have been used to learn evolve controllers and rules based upon patterns of stones for use during play. The techniques applied include Genetic Programming [10] Genetic Algorithms [32,22,12], and Neural Networks [13] These approaches have so far been less successful than the hybrid programs but have the advantage that Go knowledge does not need to be added by hand. ffl Cazenave s GoGol [8] applies learning techniques to good effect. An offline program uses introspection to prove ....

N. Richards, D. Moriarty, and R. Miikkulainen. Evolving Neural Networks to Play Go. Technical report, The University of Texas at Austin., 1997.

....parents from the population based on their fitness. 5. Apply search operators to the parents and generate offspring which form the next generation. Figure 6: A typical cycle of the evolution of architectures. Considerable research on evolving ANN architectures has been carried out in recent years [33, 42, 45, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 149, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 209, 210, 211, 212, 138, 213, 214, 215, 216, 118, 130, 127, 217, 218, 219, 220, 221, 222, 223, 128, 224, 225]. Most of the research has concentrated on the evolution of ANN topological structures. Relatively little has been done on the evolution of node transfer functions, let al..one the simultaneous evolution of both topological structures and node transfer functions. In this paper, we will analyze the ....

....Encoding Scheme Two different approaches have been taken in the direct encoding scheme. The first separates the evolution of architectures from that of connection weights [154, 153, 150, 24, 170, 169, 165, 167] The second approach evolves architectures and connection weights simultaneously [179, 180, 182, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 149, 198, 199, 200]. This section will focus on the first approach. The second approach will be discussed in Section 3.4. In the first approach, each connection of an architecture is directly specified by its binary representation [154, 153, 150, 24, 170, 169, 165, 167, 202] For example, an N Theta N matrix C = c ....

[Article contains additional citation context not shown here]

N. Richards, D. E. Moriarty, and R. Miikkulainen, "Evolving neural networks to play Go," Applied Intelligence, vol. 8, no. 1, pp. 85--96, 1998.

.... be tested against a subset of parasites, but also against the best hosts of previous generations (i.e. a hall of fame, Rosin, 1997) As the evolutionary technique we will use the SANE neuro evolution method (Moriarty and Miikkulainen, 1997) because it has been shown effective in the go domain (Richards et al. 1998). The results show that the learning speed is increased by using the co evolutionary techniques and the level of play is not limited by existing opponents. 2 THE GAME OF GO The game of go is an ancient board game which is believed to have originated in China. The rules of the game are relatively ....

....use for evaluating the performance of the evolved networks) uses it as well. In order to give the players an equal chance of winning the game, the weaker player may be given some points in advance, called komi . We will utilise komi in our experiments to make competition more even. 3 APPROACH Richards et al. 1998) evolved a go playing network for a small board with good results. However, the level of play was limited by the level of the existing program used as a sparring partner . What is the use of evolving further when you are already able to beat your opponent This problem can be addressed by ....

[Article contains additional citation context not shown here]

Richards, N., Moriarty, D., and Miikkulainen, R. (1998). Evolving neural networks to play go. Applied Intelligence, 8:85--96.

No context found.

Richards, N., Moriarty, D., and Miikkulainen, R. (in press). Evolving Neural Networks to Play Go. Applied Intelligence. Werner, G., and Dyer, M. (1991). Evolution of Communication in Artificial Organisms. Artificial Life II, 659-687. Reading, MA: Addison Wesley.

No context found.

Richards, N., Moriarty, D., and Miikkulainen, R. Evolving neural networks to play Go. Applied Intelligence, 8:85--96, 1997.

No context found.

N. Richards, D. Moriarty, and R. Miikkulainen, "Evolving neural networks to play go," Applied Intelligence, vol. 8, pp. 85--96, 1998.

No context found.

N. Richards, D.E. Moriarty, and R. Miikkulainen. Evolving neural networks to play Go. In Proceedings of the 7th International Conference on Genetic Algorithms, East Lansing, NI, USA, 1997.

No context found.

N. Richards, D.E. Moriarty, and R. Miikkulainen. Evolving neural networks to play Go. In Proceedings of the 7th International Conference on Genetic Algorithms, East Lansing, NI, USA, 1997.

No context found.

N. Richards, D. E. Moriarty, and R. Miikkulainen, "Evolving neural networks to play go," in Proc. 7th Int. Conf. Genetic Algorithms, 1997, pp. 768--775.

No context found.

N. Richards, D. Moriarty, P. McQuesten, and R. Miikkulainen. Evolving neural networks to play go. In Proceedings of the 7th International Conference on Genetic Algorithms, East Lansing, MI, 1997.

No context found.

Norman Richards, David Moriarty, Paul McQuesten, and Risto Miikkulainen. Evolving neural networks to play go. In Proceedings of the 7th International Conference on Genetic Algorithms, East Lansing, MI, 1997.

No context found.

N. Richards, D. Moriarty, and R. Miikkulainen, "Evolving neural networks to play go," Applied Intelligence, Vol. 8, pp. 85-96, 1998.

No context found.

N. Richards, D. E. Moriarty, and R. Miikkulainen, "Evolving neural networks to play Go," Appl. Intell., vol. 8, no. 1, pp. 85--96, 1998.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC