MS-TDNN with global discriminant trainings

Emilie Caillault, Christian Viard-Gaudin, Abdul Rahim Ahmad

Research output: Chapter in Book/Report/Conference proceedingConference contribution

7 Citations (Scopus)

Abstract

This article analyses the behavior of various hybrid architectures based on a multi-state neuro-markovian scheme (MS-TDNN H MM) applied to online handwriting word recognition systems. We have considered different cost functions, including maximal mutual information criteria with discriminant training and maximum likelihood estimation, to train the systems globally at the word level and also we varied the number of states from one up to three to model the basic hidden Markov models at the letter level. We report experimental results for non constrained, writer independent, word recognition obtained on the IRONOFF database.

Original languageEnglish
Title of host publicationProceedings of the Eighth International Conference on Document Analysis and Recognition
Pages856-860
Number of pages5
DOIs
Publication statusPublished - 01 Dec 2005
Event8th International Conference on Document Analysis and Recognition - Seoul, Korea, Republic of
Duration: 31 Aug 200501 Sep 2005

Publication series

NameProceedings of the International Conference on Document Analysis and Recognition, ICDAR
Volume2005
ISSN (Print)1520-5363

Other

Other8th International Conference on Document Analysis and Recognition
CountryKorea, Republic of
CitySeoul
Period31/08/0501/09/05

    Fingerprint

All Science Journal Classification (ASJC) codes

  • Computer Vision and Pattern Recognition

Cite this

Caillault, E., Viard-Gaudin, C., & Ahmad, A. R. (2005). MS-TDNN with global discriminant trainings. In Proceedings of the Eighth International Conference on Document Analysis and Recognition (pp. 856-860). [1575666] (Proceedings of the International Conference on Document Analysis and Recognition, ICDAR; Vol. 2005). https://doi.org/10.1109/ICDAR.2005.163