Design for recurrent fuzzy neural networks using MSC-MFS and PSO-MBP

A novel hybrid learning algorithm for designing a TSK-type recurrent fuzzy neural network (RFNN) is proposed in this paper. The whole designing process includes two stages, i.e., structure identification and parameter optimization. The structure identification includes mean shift clustering (MSC) and mean firing strength (MFS). The MSC is used to partition the input space and the mean firing strength (MFS) is employed to prune the redundant rule neurons. After the structure identification is performed, we adopt the PSO to adjust the free parameters of the RFNN and generate the near-optimal free parameters solution. Then, MBP is used to continue the learning process until the terminal condition is satisfied. The proposed hybrid learning algorithm achieves superior performance in learning accuracy.

[1]  Anders Krogh,et al.  Introduction to the theory of neural computation , 1994, The advanced book program.

[2]  Noureddine Zerhouni,et al.  Recurrent radial basis function network for time-series prediction , 2003 .

[3]  Cheng-Jian Lin,et al.  The Design of Neuro-Fuzzy Networks Using Particle Swarm Optimization and Recursive Singular Value Decomposition , 2006, The 2006 IEEE International Joint Conference on Neural Network Proceedings.

[4]  Kazuo Tanaka,et al.  Modeling and control of carbon monoxide concentration using a neuro-fuzzy technique , 1995, IEEE Trans. Fuzzy Syst..

[5]  C. S. George Lee,et al.  Neural fuzzy systems: a neuro-fuzzy synergism to intelligent systems , 1996 .

[6]  Ching-Hung Lee,et al.  Identification and control of dynamic systems using recurrent fuzzy neural networks , 2000, IEEE Trans. Fuzzy Syst..

[7]  T. Martin McGinnity,et al.  Design for Self-Organizing Fuzzy Neural Networks Based on Genetic Algorithms , 2006, IEEE Transactions on Fuzzy Systems.

[8]  Thomas Bäck,et al.  An Overview of Evolutionary Algorithms for Parameter Optimization , 1993, Evolutionary Computation.

[9]  M. Gupta,et al.  Approximation of discrete-time state-space trajectories using dynamic recurrent neural networks , 1995, IEEE Trans. Autom. Control..

[10]  Chia-Feng Juang,et al.  A recurrent self-organizing neural fuzzy inference network , 1997, Proceedings of 6th International Fuzzy Systems Conference.

[11]  James Kennedy,et al.  Particle swarm optimization , 1995, Proceedings of ICNN'95 - International Conference on Neural Networks.

[12]  Chia-Feng Juang,et al.  A hybrid of genetic algorithm and particle swarm optimization for recurrent network design , 2004, IEEE Transactions on Systems, Man, and Cybernetics, Part B (Cybernetics).

[13]  Shigeo Abe,et al.  Neural Networks and Fuzzy Systems , 1996, Springer US.

[14]  Jie Zhang,et al.  Recurrent neuro-fuzzy networks for nonlinear process modeling , 1999, IEEE Trans. Neural Networks.

[15]  Hsuan-Ming Feng,et al.  Self-generation RBFNs using evolutional PSO learning , 2006, Neurocomputing.

[16]  Spyros G. Tzafestas,et al.  Neural fuzzy control systems with structure and parameter learning , 1996, J. Intell. Robotic Syst..

[17]  Yuichi Nakamura,et al.  Approximation of dynamical systems by continuous time recurrent neural networks , 1993, Neural Networks.

[18]  Fei-Yue Wang,et al.  Implementing Adaptive Fuzzy Logic Controllers with Neural Networks: A Design Paradigm , 1995, J. Intell. Fuzzy Syst..

[19]  Jyh-Shing Roger Jang,et al.  ANFIS: adaptive-network-based fuzzy inference system , 1993, IEEE Trans. Syst. Man Cybern..