Optimal control with reinforcement learning using reservoir computing and Gaussian Mixture

Optimal control problems could be solved with reinforcement learning. However it is challenging to use it with continuous state and action spaces, not to speak about partially observable environments. In this paper we propose a reinforcement learning system for partially observable environments with continuous state and action spaces. The method utilizes novel machine learning methods, the Echo State Network, and the Incremental Gaussian Mixture Network.

[1]  Milton Roberto Heinen,et al.  A connectionist approach for incremental function approximation and on-line tasks , 2011 .

[2]  Christian Böhm,et al.  Independent quantization: an index compression technique for high-dimensional data spaces , 2000, Proceedings of 16th International Conference on Data Engineering (Cat. No.00CB37073).

[3]  Shie Mannor,et al.  Reinforcement learning with Gaussian processes , 2005, ICML.

[4]  András Lörincz,et al.  Reinforcement Learning with Echo State Networks , 2006, ICANN.

[5]  Herbert Jaeger,et al.  A tutorial on training recurrent neural networks , covering BPPT , RTRL , EKF and the " echo state network " approach - Semantic Scholar , 2005 .

[6]  Paulo Martins Engel,et al.  An Incremental Probabilistic Neural Network for Regression and Reinforcement Learning Tasks , 2010, ICANN.

[7]  Sebastian Thrun,et al.  Issues in Using Function Approximation for Reinforcement Learning , 1999 .

[8]  Richard S. Sutton,et al.  Reinforcement Learning: An Introduction , 1998, IEEE Trans. Neural Networks.

[9]  Liming Xiang,et al.  Kernel-Based Reinforcement Learning , 2006, ICIC.

[10]  Hans-Peter Kriegel,et al.  The X-tree : An Index Structure for High-Dimensional Data , 2001, VLDB.

[11]  Alejandro Agostini,et al.  Reinforcement Learning with a Gaussian mixture model , 2010, The 2010 International Joint Conference on Neural Networks (IJCNN).

[12]  Carl E. Rasmussen,et al.  Gaussian process dynamic programming , 2009, Neurocomputing.

[13]  Keith A. Bush.  An echo state model of non-markovian reinforcement learning , 2007 .