How to deal with uncertainty and variability: experience and solutions

Authors:
Hiromichi Fujisawa
Affiliations:
Central Research Laboratory, Hitachi, Ltd., Tokyo, Japan
Venue:
SACH'06 Proceedings of the 2006 conference on Arabic and Chinese handwriting recognition
Year:
2006

Citing 8
Cited 1

A Dynamic Programming Approach to Sequential Pattern Recognition

IEEE Transactions on Pattern Analysis and Machine Intelligence - Special memorial issue for Professor King-Sun Fu
Decision Combination in Multiple Classifier Systems

IEEE Transactions on Pattern Analysis and Machine Intelligence
Off-Line, Handwritten Numeral Recognition by Perturbation Method

IEEE Transactions on Pattern Analysis and Machine Intelligence
Lexicon-Driven Segmentation and Recognition of Handwritten Character Strings for Japanese Address Reading

IEEE Transactions on Pattern Analysis and Machine Intelligence
Lexicon-driven word recognition

ICDAR '95 Proceedings of the Third International Conference on Document Analysis and Recognition (Volume 2) - Volume 2
Pattern Classification (2nd Edition)

Pattern Classification (2nd Edition)
Address-Block Extraction by Bayesian Rule

ICPR '04 Proceedings of the Pattern Recognition, 17th International Conference on (ICPR'04) Volume 2 - Volume 02
Spiral Recognition Methodology and Its Application for Recognition of Chinese Bank Checks

IWFHR '04 Proceedings of the Ninth International Workshop on Frontiers in Handwriting Recognition

Forty years of research in character and document recognition-an industrial perspective

Pattern Recognition

Quantified Score

Hi-index	0.00

Visualization

Abstract

Uncertainty and variability are two of the most important concepts at the center of pattern recognition. It is especially true when patterns to be recognized are complex in nature and not controlled by any artificial constraints. Handwritten postal address recognition is one such case. This paper presents five principles of dealing with uncertainty and variability, and discusses how to decompose the complex recognition task into manageable sub-tasks. When applicable, block diagrams will clarify the structure of various recognition components. This paper also presents implementation of those principles into real recognition engines. It will demonstrate that high accuracy and robustness of a recognition system, which relates to uncertainty and variability, respectively, can occur only with comprehensive approaches.