Abstract
We have developed the notion of lexicon density as the true metric to measure expected recognizer accuracy. This metric has a variety of applications, among them evaluation of recognition results, static or dynamic recognizer selection, or dynamic combination of recognizers. We show that the performance of word recognizers increases as lexicon density decreases and that the relationship between the performance and lexicon density is independent of lexicon size. Our claims are supported by extensive experimental validation data.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
Reference
L. R. Bahl, F. Jelinek, and R. L. Mercer. A maximum likelihood approach to continuous speech recognition. IEEE Transactions on Pattern Analysis and Machine Intelligence, 5(2), Mar. 1983.
G. Kim and V. Govindaraju. Bank check recognition using cross validation between legal and courtesy amounts. International Journal of Pattern recognition and Artificial Intelligence, 11(4): 657–674, 1997.
G. Kim and V. Govindaraju. A lexicon driven approach to handwritten word recognition for real-time applications. IEEE Transactions on Pattern Analysis and Machine Intelligence, 19(4): 366–379, Apr. 1997.
P. SlavĂk. Slice distance. Technical Report 2000-04, Department of Computer Science, SUNY at Buffalo, Apr. 2000. http://www.cse.buffalo.edu/tech-reports/.
R. A. Wagner and M. J. Fischer. The string-to-string correction problem. J. ACM, 2(1): 168–173, Jan. 1974.
Author information
Authors and Affiliations
Rights and permissions
Copyright information
© 2000 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
SlavĂk, P., Govindaraju, V. (2000). Use of Lexicon Density in Evaluating Word Recognizers. In: Multiple Classifier Systems. MCS 2000. Lecture Notes in Computer Science, vol 1857. Springer, Berlin, Heidelberg. https://doi.org/10.1007/3-540-45014-9_30
Download citation
DOI: https://doi.org/10.1007/3-540-45014-9_30
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-67704-8
Online ISBN: 978-3-540-45014-6
eBook Packages: Springer Book Archive