Robustness through prior knowledge: Using explanation-based learning to distinguish handwritten Chinese characters

Qiang Sun, Li Lun Wang, Shiau Hong Lim, Gerald DeJong

Research output: Contribution to journalArticlepeer-review

Abstract

Handwritten Chinese character recognition is difficult due to the unstructured and noisy nature of its training examples. There are often too few training examples for a statistical learner like SVM to overcome the noise and extract useful information reliably. Existing prior domain knowledge represents a valuable source of information for classifying handwritten characters. Explanation-based learning (EBL) provides a way to incorporating prior domain knowledge into the learner. The dynamic bias formed by the interaction of domain knowledge with training examples can yield solution knowledge of potential higher quality. Two EBL approaches, one that uses a special feature kernel function in SVM, the other uses a conventional kernel for the SVM but provides additional preference in choosing the classification hyperplane, are reported.

Original languageEnglish (US)
Pages (from-to)175-186
Number of pages12
JournalInternational Journal on Document Analysis and Recognition
Volume10
Issue number3-4
DOIs
StatePublished - Dec 1 2007

Keywords

  • Characterr ecognition
  • Domain knowledge
  • Explanation-based learning
  • Machine learning
  • Support vector machine

ASJC Scopus subject areas

  • Software
  • Computer Vision and Pattern Recognition
  • Computer Science Applications

Fingerprint

Dive into the research topics of 'Robustness through prior knowledge: Using explanation-based learning to distinguish handwritten Chinese characters'. Together they form a unique fingerprint.

Cite this