SOM-based information maximization to improve and interpret multi-layered neural networks: From information reduction to information augmentation approach to create new information
作者：
Highlights：
• Necessity of information augmentation in neural networks.
• New interpretation of information-theoretic methods in neural networks.
• New interpretation of SOM in creating new knowledge.
• New learning methods for training multi-layered neural networks and for coping with the vanishing information problem.
• Demonstration of improved generalization with excessive information.
摘要
•Necessity of information augmentation in neural networks.•New interpretation of information-theoretic methods in neural networks.•New interpretation of SOM in creating new knowledge.•New learning methods for training multi-layered neural networks and for coping with the vanishing information problem.•Demonstration of improved generalization with excessive information.
论文关键词：Information maximization,Vanishing information,Pre-training,SOM,Generalization,Interpretation
论文评审过程：Received 3 April 2018, Revised 26 November 2018, Accepted 24 January 2019, Available online 1 February 2019, Version of Record 15 February 2019.
论文官网地址：https://doi.org/10.1016/j.eswa.2019.01.056
