Neural input selection-A fast model-based approach

Authors:
Kang Li;Jian-Xun Peng
Affiliations:
School of Electrical & Electronic Engineering, Queen's University Belfast, Ashby Building, Stranmillis Road, Belfast BT9 5AH, UK;School of Electrical & Electronic Engineering, Queen's University Belfast, Ashby Building, Stranmillis Road, Belfast BT9 5AH, UK
Venue:
Neurocomputing
Year:
2007

Citing 8
Cited 4

System identification: theory for the user

System identification: theory for the user
Optimal brain damage

Advances in neural information processing systems 2
Model selection in neural networks

Neural Networks
Pruning from adaptive regularization

Neural Computation
Identification and control of dynamical systems using neural networks

IEEE Transactions on Neural Networks
A generalized growing and pruning RBF (GGAP-RBF) neural network for function approximation

IEEE Transactions on Neural Networks
Pruning algorithms-a survey

IEEE Transactions on Neural Networks
Pruning recurrent neural networks for improved generalization performance

IEEE Transactions on Neural Networks

Combined input variable selection and model complexity control for nonlinear regression

Pattern Recognition Letters
Simultaneous input variable and basis function selection for RBF networks

Neurocomputing
A fast multi-output RBF neural network construction method

Neurocomputing
An efficient feature selection method for mobile devices with application to activity recognition

Neurocomputing

Quantified Score

Hi-index	0.01

Visualization

Abstract

Neural input selection is an important stage in neural network configuration. For neural modeling and control of nonlinear dynamic systems, the inputs to the neural networks may include any system variable of interest with various time lags. To choose a set of significant inputs is a combinational problem, and the selection procedure can be very time consuming. In this paper, a model-based neural input selection method is proposed. Essentially, the neural input selection is transformed into the problem of identifying the significant terms for a linear-in-the-parameters model. A fast method is then proposed to identify significant nonlinear terms or functions, from which the neural inputs are grouped and selected. Both theoretic analysis and simulation examples demonstrate the effectiveness and efficiency of the proposed model-based approach.