Handwritten Chinese Character Recognition by Convolutional Neural Network and Similarity Ranking
Handwritten Chinese Character Recognition by Convolutional Neural Network and Similarity Ranking
About this item
Full title
Author / Creator
Publisher
Ithaca: Cornell University Library, arXiv.org
Journal title
Language
English
Formats
Publication information
Publisher
Ithaca: Cornell University Library, arXiv.org
Subjects
More information
Scope and Contents
Contents
Convolution Neural Networks (CNN) have recently achieved state-of-the art performance on handwritten Chinese character recognition (HCCR). However, most of CNN models employ the SoftMax activation function and minimize cross entropy loss, which may cause loss of inter-class information. To cope with this problem, we propose to combine cross entropy...
Alternative Titles
Full title
Handwritten Chinese Character Recognition by Convolutional Neural Network and Similarity Ranking
Authors, Artists and Contributors
Author / Creator
Identifiers
Primary Identifiers
Record Identifier
TN_cdi_proquest_journals_2283170587
Permalink
https://devfeature-collection.sl.nsw.gov.au/record/TN_cdi_proquest_journals_2283170587
Other Identifiers
E-ISSN
2331-8422
