skip to main content
Show Results with:

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

Yang, Weixin et al.

Pattern recognition -- Elsevier Science -- Volume: 58 C; (pages 190-203) -- 2016

Online access

  • Title:
    DropSample: A new training method to enhance deep convolutional neural networks for large-scale unconstrained handwritten Chinese character recognition
  • Author: Yang, Weixin;
    Jin, Lianwen;
    Tao, Dacheng;
    Xie, Zecheng;
    Feng, Ziyong
  • Found In: Pattern recognition. Volume 58:Number C(2016); 201610; 190-203
  • Journal Title: Pattern recognition
  • Subjects: Pattern perception; Perception des structures Périodiques; Patroonherkenning; LCSH: Pattern perception; Dewey: 006.4
  • Rights: Licensed
  • Publication Details: Elsevier Science
  • Abstract: AbstractInspired by the theory of Leitner׳s learning box from the field of psychology, we proposeDropSample, a new method for training deep convolutional neural networks (DCNNs), and apply it to large-scale online handwritten Chinese character recognition (HCCR). According to the principle ofDropSample, each training sample is associated with a quota function that is dynamically adjusted on the basis of the classification confidence given by the DCNN softmax output. After a learning iteration, samples with low confidence will have a higher frequency of being selected as training data; in contrast, well-trained and well-recognized samples with very high confidence will have a lower frequency of being involved in the ongoing training and can be gradually eliminated. As a result, the learning process becomes more efficient as it progresses. Furthermore, we investigate the use of domain-specific knowledge to enhance the performance of DCNN by adding a domain knowledge layer before the traditional CNN. By adoptingDropSampletogether with different types of domain-specific knowledge, the accuracy of HCCR can be improved efficiently. Experiments on the CASIA-OLHDWB 1.0, CASIA-OLHWDB 1.1, and ICDAR 2013 online HCCR competition datasets yield outstanding recognition rates of 97.33%, 97.06%, and 97.51% respectively, all of which are significantly better than the previous best results reported in the literature.HighlightsWe propose a novel and efficient training method for CNN on large-scale data.DropSampleadaptively 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.
  • Identifier: Journal ISSN: 0031-3203
  • Publication Date: 2016
  • Physical Description: Electronic
  • Shelfmark(s): ELD Digital store
  • UIN: ETOCvdc_100040981074.0x000001

Searching Remote Databases, Please Wait