Adaptive Kernel Leaning Networks with Application to Nonlinear System Identification
By kernelizing the traditional least-square based identification method, an adaptive kernel learning (AKL) network is proposed for nonlinear process modeling, which utilizes kernel mapping and geometric angle to build the network topology adaptively. The generalization ability of AKL network is controlled by introducing a regularized optimization function. Two forms of learning strategies are addressed and their corresponding recursive algorithms are derived. Numerical simulations show this simple AKL networks can learn the process nonlinearities with very small samples, and has excellent modeling performance in both the deterministic and stochastic environments.
KeywordsGeneralization Ability High Dimensional Feature Space Adaptive Neural Network Quadratic Cost Function Nonlinear System Identification
Unable to display preview. Download preview PDF.
- 2.Ljung, L.: System Identification: Theory for the User, 2nd edn. Prentice-Hall, New Jersey (1999)Google Scholar
- 4.Lixin, W.: Adaptive Fuzzy Systems and Control: Design and Stability Analysis. Prentice-Hall, New Jersey (1994)Google Scholar
- 6.Schölkopf, B., Smola, A.J.: Learning with Kernels: Support Vector Machines, Regularization, Optimization, and Beyond. The MIT Press, Cambridge (2002)Google Scholar
- 7.Engel, Y., Mannor, S., Meir, R.: Kernel Recursive Least Squares (2003), http://www.ee.technion.ac.il/~rmeir/Publications/KrlsReport.pdf
- 8.Csató, L., Opper, M.: Sparse representation for Gaussian process models. In: Leen, T.K., Dietterich, T.G., Tresp, V. (eds.) Advances in Neural Information Processing Systems, vol. 13, pp. 444–450. MIT Press, Cambridge (2001)Google Scholar
- 9.Franc, V., Hlavac, V.: Training set approximation for kernel methods. In: Proceedings of the 8th Computer Vision Winter Workshop, Prague, Czech Republic, Czech Pattern Recognition Society, pp. 121–126 (2003)Google Scholar