The C-loss function for pattern classification
作者：
Highlights：
• We propose a new classification method.
• The method is based on the recently proposed similarity measure called Correntropy.
• The method is robust against overfitting and less sensitive to system parameters.
• We obtain better performance as compared to several conventional classifiers.
摘要
Highlights•We propose a new classification method.•The method is based on the recently proposed similarity measure called Correntropy.•The method is robust against overfitting and less sensitive to system parameters.•We obtain better performance as compared to several conventional classifiers.
论文关键词：Classification,Correntropy,Neural network,Loss function,Backprojection
论文评审过程：Received 3 October 2012, Revised 16 July 2013, Accepted 24 July 2013, Available online 3 August 2013.
论文官网地址：https://doi.org/10.1016/j.patcog.2013.07.017
