DropSample: A new training method to enhance deep convolutional neural networks for large-scale unconstrained handwritten Chinese character recognition

作者：

Highlights：

• We propose a novel and efficient training method for CNN on large-scale data.

• DropSample adaptively selects training samples and is robust to noisy data.

• The incorporation of domain-specific knowledge enhances the performance of CNN.

• New state-of-the-art results are reported on 3 online handwritten Chinese character datasets.

摘要

•We propose a novel and efficient training method for CNN on large-scale data.•DropSample adaptively selects training samples and is robust to noisy data.•The incorporation of domain-specific knowledge enhances the performance of CNN.•New state-of-the-art results are reported on 3 online handwritten Chinese character datasets.

论文关键词：Convolutional neural network,Deep learning,Handwritten character recognition,Domain-specific knowledge

论文评审过程：Received 24 March 2015, Revised 20 March 2016, Accepted 13 April 2016, Available online 23 April 2016, Version of Record 26 May 2016.

论文官网地址：https://doi.org/10.1016/j.patcog.2016.04.007