Cross-validation based weights and structure determination of Chebyshev-polynomial neural networks for pattern classification

作者：

Highlights：

• Two neural networks are constructed and investigated for pattern classification.

• The proposed neural networks are of low computational complexity.

• Two proposed WASD algorithms can obtain proper structures of neural networks.

• The neural networks with WASD achieve high classification accuracy.

• The MOCPNN with WASD achieves strong robustness on classification.

摘要

Highlights•Two neural networks are constructed and investigated for pattern classification.•The proposed neural networks are of low computational complexity.•Two proposed WASD algorithms can obtain proper structures of neural networks.•The neural networks with WASD achieve high classification accuracy.•The MOCPNN with WASD achieves strong robustness on classification.

论文关键词：Cross validation,Chebyshev polynomial,Neural network,Pattern classification,Robustness

论文评审过程：Received 28 June 2013, Revised 8 April 2014, Accepted 29 April 2014, Available online 14 May 2014.

论文官网地址：https://doi.org/10.1016/j.patcog.2014.04.026