摘要: Neuroevolution, i.e. evolving artificial neural networks with genetic algorithms, has been highly effective in reinforcement learning tasks, particularly those hidden state information. An important question neuroevolution is how to gain an advantage from network topologies along weights. We present a method, NeuroEvolution of Augmenting Topologies (NEAT) that outperforms the best fixed-topology methods on challenging benchmark task. claim increased efficiency due (1) employing principled method crossover different topologies, (2) protecting structural innovation using speciation, and (3) incrementally growing minimal structure. test this through series ablation studies demonstrate each component necessary system as whole other. What results significantly faster learning. NEAT also contribution GAs because it shows possible for evolution both optimize complexify solutions simultaneously, making evolve increasingly complex over time, thereby strengthening analogy biological evolution.
sci-hub.se 参考文章(20)
Heinz Mühlenbein, Byoung-Tak Zhang, Evolving Optimal Neural Networks Using Genetic Algorithms with Occam's Razor. Complex Systems. ,vol. 7, ,(1993)
Alexis P. Wieland, Evolving Controls for Unstable Systems Connectionist Models#R##N#Proceedings of the 1990 Summer School. pp. 91- 102 ,(1991) , 10.1016/B978-1-4832-1448-1.50015-9
Frederic Gruau, Darrell Whitley, Larry Pyeatt, None, A comparison between cellular encoding and direct encoding for genetic neural networks Proceedings of the 1st annual conference on genetic programming. pp. 81- 89 ,(1996)
David Eric Moriarty, Symbiotic Evolution of Neural Networks in Sequential Decision Tasks University of Texas at Austin. ,(1997)
Samir W. Mahfoud, Niching methods for genetic algorithms University of Illinois at Urbana-Champaign. ,(1996)
João Carlos Figueira Pujol, Riccardo Poli, Evolving the Topology and the Weights of Neural Networks Using a Dual Representation Applied Intelligence. ,vol. 8, pp. 73- 84 ,(1998) , 10.1023/A:1008272615525
N. Saravanan, D.B. Fogel, Evolving neural control systems IEEE Intelligent Systems. ,vol. 10, pp. 23- 27 ,(1995) , 10.1109/64.393139
Darrell Whitley, Stephen Dominic, Rajarshi Das, Charles W. Anderson, Genetic Reinforcement Learning for Neurocontrol Problems Machine Learning. ,vol. 13, pp. 259- 284 ,(1993) , 10.1007/BF00993045
Andrew G. Barto, Richard S. Sutton, Charles W. Anderson, Neuronlike adaptive elements that can solve difficult learning control problems systems man and cybernetics. ,vol. 13, pp. 834- 846 ,(1983) , 10.1109/TSMC.1983.6313077
Faustino Gomez, Risto Miikkulainen, Incremental Evolution of Complex General Behavior Adaptive Behavior. ,vol. 5, pp. 317- 342 ,(1997) , 10.1177/105971239700500305