Writer adaptation of a HMM handwriting recognition system

Author

Senior, Andrew ; Nathan, Krishna

Author_Institution

IBM Thomas J. Watson Res. Center, Yorktown Heights, NY, USA

Volume

2

fYear

1997

fDate

21-24 Apr 1997

Firstpage

1447

Abstract

This paper describes a scheme to adapt the parameters of a tied-mixture, hidden Markov model, on-line handwriting recognition system to improve performance on new writers´ handwriting. The means and variances of the distributions are adapted using the maximum likelihood linear regression technique. Experiments are performed with a number of new writers in both supervised and unsupervised modes. Adaptation on data quantities as small as 5 words is found to result in models with 6% lower error rate than the writer independent model

Keywords

adaptive estimation; handwriting recognition; hidden Markov models; maximum likelihood estimation; HMM handwriting recognition system; distribution means; distribution variances; error rate; maximum likelihood linear regression technique; models; performance; supervised modes; tied-mixture hidden Markov model; unsupervised modes; writer adaptation; writer independent model; Error analysis; Handwriting recognition; Hidden Markov models; Maximum likelihood estimation; Maximum likelihood linear regression; Probability distribution; Shape; Speech recognition; Statistics; Writing;

fLanguage

English

Publisher

ieee

Conference_Titel

Acoustics, Speech, and Signal Processing, 1997. ICASSP-97., 1997 IEEE International Conference on

Conference_Location

Munich

ISSN

1520-6149

Print_ISBN

0-8186-7919-0

Type

conf

DOI

10.1109/ICASSP.1997.596221

Filename

596221