Watching pattern distribution via massive character recognition

Seiichi Uchida, Wenjie Cai, Akira Yoshida, Yaokai Feng

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

Abstract

The purpose of this paper is to analyze how image patterns distribute inside their feature space. For this purpose, 832,612 manually ground-truthed handwritten digit patterns are used. Use of character patterns instead of general visual object patterns is very essential for our purpose. First, since there are only 10 classes for digits, it is possible to have an enough number of patterns per class. Second, since the feature space of small binary character images is rather compact, it is easier to observe the precise pattern distribution with a fixed number of patterns. Third, the classes of character patterns can be defined far more clearly than visual objects. Through nearest neighbor analysis on 832, 612 patterns, their distribution in the 32 x 32 binary feature space is observed quantitatively and qualitatively. For example, the visual similarity of nearest neighbors and the existence of outliers, which are surrounded by patterns from different classes, are observed.

Original languageEnglish
Title of host publication2011 IEEE International Workshop on Machine Learning for Signal Processing - Proceedings of MLSP 2011
DOIs
Publication statusPublished - Dec 5 2011
Event21st IEEE International Workshop on Machine Learning for Signal Processing, MLSP 2011 - Beijing, China
Duration: Sep 18 2011Sep 21 2011

Publication series

NameIEEE International Workshop on Machine Learning for Signal Processing

Other

Other21st IEEE International Workshop on Machine Learning for Signal Processing, MLSP 2011
CountryChina
CityBeijing
Period9/18/119/21/11

Fingerprint

Character recognition

All Science Journal Classification (ASJC) codes

  • Human-Computer Interaction
  • Signal Processing

Cite this

Uchida, S., Cai, W., Yoshida, A., & Feng, Y. (2011). Watching pattern distribution via massive character recognition. In 2011 IEEE International Workshop on Machine Learning for Signal Processing - Proceedings of MLSP 2011 [6064640] (IEEE International Workshop on Machine Learning for Signal Processing). https://doi.org/10.1109/MLSP.2011.6064640

Watching pattern distribution via massive character recognition. / Uchida, Seiichi; Cai, Wenjie; Yoshida, Akira; Feng, Yaokai.

2011 IEEE International Workshop on Machine Learning for Signal Processing - Proceedings of MLSP 2011. 2011. 6064640 (IEEE International Workshop on Machine Learning for Signal Processing).

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

Uchida, S, Cai, W, Yoshida, A & Feng, Y 2011, Watching pattern distribution via massive character recognition. in 2011 IEEE International Workshop on Machine Learning for Signal Processing - Proceedings of MLSP 2011., 6064640, IEEE International Workshop on Machine Learning for Signal Processing, 21st IEEE International Workshop on Machine Learning for Signal Processing, MLSP 2011, Beijing, China, 9/18/11. https://doi.org/10.1109/MLSP.2011.6064640
Uchida S, Cai W, Yoshida A, Feng Y. Watching pattern distribution via massive character recognition. In 2011 IEEE International Workshop on Machine Learning for Signal Processing - Proceedings of MLSP 2011. 2011. 6064640. (IEEE International Workshop on Machine Learning for Signal Processing). https://doi.org/10.1109/MLSP.2011.6064640
Uchida, Seiichi ; Cai, Wenjie ; Yoshida, Akira ; Feng, Yaokai. / Watching pattern distribution via massive character recognition. 2011 IEEE International Workshop on Machine Learning for Signal Processing - Proceedings of MLSP 2011. 2011. (IEEE International Workshop on Machine Learning for Signal Processing).
@inproceedings{db624e7908cd46be8015f7dd9e8f347c,
title = "Watching pattern distribution via massive character recognition",
abstract = "The purpose of this paper is to analyze how image patterns distribute inside their feature space. For this purpose, 832,612 manually ground-truthed handwritten digit patterns are used. Use of character patterns instead of general visual object patterns is very essential for our purpose. First, since there are only 10 classes for digits, it is possible to have an enough number of patterns per class. Second, since the feature space of small binary character images is rather compact, it is easier to observe the precise pattern distribution with a fixed number of patterns. Third, the classes of character patterns can be defined far more clearly than visual objects. Through nearest neighbor analysis on 832, 612 patterns, their distribution in the 32 x 32 binary feature space is observed quantitatively and qualitatively. For example, the visual similarity of nearest neighbors and the existence of outliers, which are surrounded by patterns from different classes, are observed.",
author = "Seiichi Uchida and Wenjie Cai and Akira Yoshida and Yaokai Feng",
year = "2011",
month = "12",
day = "5",
doi = "10.1109/MLSP.2011.6064640",
language = "English",
isbn = "9781457716232",
series = "IEEE International Workshop on Machine Learning for Signal Processing",
booktitle = "2011 IEEE International Workshop on Machine Learning for Signal Processing - Proceedings of MLSP 2011",

}

TY - GEN

T1 - Watching pattern distribution via massive character recognition

AU - Uchida, Seiichi

AU - Cai, Wenjie

AU - Yoshida, Akira

AU - Feng, Yaokai

PY - 2011/12/5

Y1 - 2011/12/5

N2 - The purpose of this paper is to analyze how image patterns distribute inside their feature space. For this purpose, 832,612 manually ground-truthed handwritten digit patterns are used. Use of character patterns instead of general visual object patterns is very essential for our purpose. First, since there are only 10 classes for digits, it is possible to have an enough number of patterns per class. Second, since the feature space of small binary character images is rather compact, it is easier to observe the precise pattern distribution with a fixed number of patterns. Third, the classes of character patterns can be defined far more clearly than visual objects. Through nearest neighbor analysis on 832, 612 patterns, their distribution in the 32 x 32 binary feature space is observed quantitatively and qualitatively. For example, the visual similarity of nearest neighbors and the existence of outliers, which are surrounded by patterns from different classes, are observed.

AB - The purpose of this paper is to analyze how image patterns distribute inside their feature space. For this purpose, 832,612 manually ground-truthed handwritten digit patterns are used. Use of character patterns instead of general visual object patterns is very essential for our purpose. First, since there are only 10 classes for digits, it is possible to have an enough number of patterns per class. Second, since the feature space of small binary character images is rather compact, it is easier to observe the precise pattern distribution with a fixed number of patterns. Third, the classes of character patterns can be defined far more clearly than visual objects. Through nearest neighbor analysis on 832, 612 patterns, their distribution in the 32 x 32 binary feature space is observed quantitatively and qualitatively. For example, the visual similarity of nearest neighbors and the existence of outliers, which are surrounded by patterns from different classes, are observed.

UR - http://www.scopus.com/inward/record.url?scp=82455212607&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=82455212607&partnerID=8YFLogxK

U2 - 10.1109/MLSP.2011.6064640

DO - 10.1109/MLSP.2011.6064640

M3 - Conference contribution

AN - SCOPUS:82455212607

SN - 9781457716232

T3 - IEEE International Workshop on Machine Learning for Signal Processing

BT - 2011 IEEE International Workshop on Machine Learning for Signal Processing - Proceedings of MLSP 2011

ER -