An Estimator Update Scheme for Large Teams of Learning Automata
Learning Automata are stochastic decision-making machines that have been widely used in classification, control, and network routing, between others. Despite their versatility, one of the main drawbacks of these models is the low convergence rate of the learning rules used for the training. Estimator algorithms such as Pursuit schemes help to overcome this limitation, although they require a high computer memory cost for their operation. This fact becomes a serious inconvenient when a large set of learning automata collaborate in a team to solve a concrete task, since the memory requirements of these algorithms increases exponentially. In these cases, Pursuit algorithms are ineffective due to memory overflow.
In this work, we address this problem and we propose an estimator algorithm that can be used to train large teams of Learning Automata. The approach uses a similar strategy to Tabu Search algorithms to manage long and short term memory, in order to reduce the memory requirements. The method is applied in classic permutation problems as a test-bed.
KeywordsEstimator Algorithm Term Memory Quadratic Assignment Problem Learn Automaton Learn Automaton
Unable to display preview. Download preview PDF.
- 5.Glover, F., Laguna, M.: Tabu search, pp. 70–150 (1993)Google Scholar
- 6.Lukac, K., Lukac, Z., Tkalic, M.: Behaviour of f learning automata as multicriteria routing agents in connection oriented networks. In: The 12th IEEE International Conference on Fuzzy Systems, FUZZ 2003, vol. 1, pp. 296–301 (2003)Google Scholar
- 7.Narendra, K.S., Thathachar, M.A.L.: Learning automata: an introduction. Prentice-Hall, Inc., Upper Saddle River (1989)Google Scholar
- 9.Oommen, B., Agache, M.: A comparison of continuous and discretized pursuit learning schemes. In: Proceedings of 1999 IEEE International Conference on Systems, Man, and Cybernetics, SMC 1999, vol. 4, pp. 1061–1067 (1999)Google Scholar
- 12.Peeters, M., Verbeeck, K., Nowé, A.: Solving multi-stage games with hierarchical learning automata that bootstrap. In: Tuyls, K., Nowe, A., Guessoum, Z., Kudenko, D. (eds.) ALAMAS 2005, ALAMAS 2006, and ALAMAS 2007. LNCS (LNAI), vol. 4865, pp. 169–187. Springer, Heidelberg (2008)CrossRefGoogle Scholar
- 18.Torkestani, J., Meybodi, M.: Graph coloring problem based on learning automata. In: International Conference on Information Management and Engineering, ICIME 2009, pp. 718–722 (2009)Google Scholar
- 19.Torkestani, J., Meybodi, M.: Solving the minimum spanning tree problem in stochastic graphs using learning automata. In: International Conference on Information Management and Engineering, ICIME 2009, pp. 643–647 (2009)Google Scholar