Three-phase strategy for the OSD learning method in RBF neural networks

School of Engineering, Tarbiat Modares University, P.O. Box 14115-179, Tehran, Iran
Neurocomputing (Impact Factor: 2.01). 03/2009; 72(7-9):1797-1802. DOI: 10.1016/j.neucom.2008.05.011
Source: DBLP

ABSTRACT This paper presents a novel approach in learning algorithms commonly used for training radial basis function (RBF) neural networks. This approach could be used in applications that need real-time capabilities for retraining RBF neural networks. The proposed method is a Three-Phase Learning Algorithm that optimizes the functionality of the Optimum Steepest Decent (OSD) learning method. This methodology focuses to attain greater precision in initializing the center and width of RBF units. An RBF neural network with well-adjusted RBF units in the train process will result in better performance in network response. This method is proposed to reach better performance for RBF neural networks in fewer train iterations, which is the critical issue in real-time applications. Comparing results employing different learning strategies shows interesting outcomes as have come out in this paper.

1 Follower
  • [Show abstract] [Hide abstract]
    ABSTRACT: In this paper a new methodology for training radial basis function (RBF) neural networks is introduced and examined. This novel approach, called Fuzzy-OSD, could be used in applications, which need real-time capabilities for retraining neural networks. The proposed method uses fuzzy clustering in order to improve the functionality of the Optimum Steepest Descent (OSD) learning algorithm. This improvement is due to initialization of RBF units more precisely using fuzzy C-Means clustering algorithm that results in producing better and the same network response in different retraining attempts. In addition, adjusting RBF units in the network with great accuracy will result in better performance in fewer train iterations, which is essential when fast retraining of the network is needed, especially in the real-time systems. We employed this new method in an online radar pulse classification system, which needs quick retraining of the network once new unseen emitters detected. Having compared result of applying the new algorithm and Three-Phase OSD method to benchmark problems from Proben1 database and also using them in our system, we achieved improvement in the results as presented in this paper.
    Applied Soft Computing 09/2013; 13(9):3831-3838. DOI:10.1016/j.asoc.2013.04.021 · 2.68 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Parallel robotic manipulators have a complicated dynamic model due to the presence of multi-closed-loop chains and singularities. Therefore, the control of them is a challenging and difficult task. In this paper, a novel adaptive tracking controller is proposed for parallel robotic manipulators based on fully tuned radial basis function networks (RBFNs). For developing the controller, a dynamic model of a general parallel manipulator is developed based on D׳Alembert principle and principle of virtual work. RBFNs are utilized to adaptively compensate for the modeling uncertainties, frictional terms and external disturbances of the control system. The adaptation laws for the RBFNs are derived to adjust on-line the output weights and both the centers and variances of Gaussian functions. The stability of the closed-loop system is ensured by using the Lyapunov method. Finally, a simulation example is conducted for a 2 degree of freedom (DOF) parallel manipulator to illustrate the effectiveness of the proposed controller.
    Neurocomputing 08/2014; 137:12–23. DOI:10.1016/j.neucom.2013.04.056 · 2.01 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: This paper presents a new compact hardware implementation of pulse mode Radial Basis Function Neural Network (RBFNN) with on-chip learning capacity. Since, hardware on-chip learning is a difficult issue, this work deals a hybrid process based into two stages. To update the centers positions of the radial activation functions, we apply the K-means algorithm, while to modify the connection weights; we used the back-propagation algorithm. The hardware implementation steeps of the whole network are given in details. The corresponding design was validated and implemented into the FPGA platform. To ensure the efficiency of the proposed design, we consider edge detection operation, which is a very important step in image processing. Experiential results show good approximation features and effective generalization test.
    Computer and Information Technology (WCCIT), 2013 World Congress on; 01/2013


Available from
May 19, 2014