A Multi-Class SVM Classification System Based on Methods of Self-Learning and Error Filtering

碩士 === 國立中正大學 === 資訊工程所 === 95 === This study uses Support Vector Machine (SVM) to deal with multi-class classification. It coordinates several data retrieving techniques including word segmentation, term weighting and feature extraction to achieve Chinese text classification. To improve system accu...

Full description

Bibliographic Details
Main Authors:	Chih-Hsiung Huang, 黃志雄
Other Authors:	SingLing Lee
Format:	Others
Language:	en_US
Published:	2007
Online Access:	http://ndltd.ncl.edu.tw/handle/21380011516814819107

id	ndltd-TW-095CCU05392073
record_format	oai_dc
spelling	ndltd-TW-095CCU053920732015-10-13T11:31:38Z http://ndltd.ncl.edu.tw/handle/21380011516814819107 A Multi-Class SVM Classification System Based on Methods of Self-Learning and Error Filtering 利用自動學習與錯誤過濾方法的多類別支援向量機分類系統 Chih-Hsiung Huang 黃志雄碩士國立中正大學資訊工程所 95 This study uses Support Vector Machine (SVM) to deal with multi-class classification. It coordinates several data retrieving techniques including word segmentation, term weighting and feature extraction to achieve Chinese text classification. To improve system accuracy, two methods, self-learning and error filtering are proposed. The method of self-learning uses misclassified documents to retrain classification system. The method of error filtering filters out possibly misclassified documents by analyzing the decision values from SVM. The experiment results on real-world data set shows the accuracy of basic SVM classification system is about 79% and the accuracy of improved SVM classification system can reach 83%. SingLing Lee 李新林 2007 學位論文 ; thesis 43 en_US
collection	NDLTD
language	en_US
format	Others
sources	NDLTD
description	碩士 === 國立中正大學 === 資訊工程所 === 95 === This study uses Support Vector Machine (SVM) to deal with multi-class classification. It coordinates several data retrieving techniques including word segmentation, term weighting and feature extraction to achieve Chinese text classification. To improve system accuracy, two methods, self-learning and error filtering are proposed. The method of self-learning uses misclassified documents to retrain classification system. The method of error filtering filters out possibly misclassified documents by analyzing the decision values from SVM. The experiment results on real-world data set shows the accuracy of basic SVM classification system is about 79% and the accuracy of improved SVM classification system can reach 83%.
author2	SingLing Lee
author_facet	SingLing Lee Chih-Hsiung Huang 黃志雄
author	Chih-Hsiung Huang 黃志雄
spellingShingle	Chih-Hsiung Huang 黃志雄 A Multi-Class SVM Classification System Based on Methods of Self-Learning and Error Filtering
author_sort	Chih-Hsiung Huang
title	A Multi-Class SVM Classification System Based on Methods of Self-Learning and Error Filtering
title_short	A Multi-Class SVM Classification System Based on Methods of Self-Learning and Error Filtering
title_full	A Multi-Class SVM Classification System Based on Methods of Self-Learning and Error Filtering
title_fullStr	A Multi-Class SVM Classification System Based on Methods of Self-Learning and Error Filtering
title_full_unstemmed	A Multi-Class SVM Classification System Based on Methods of Self-Learning and Error Filtering
title_sort	multi-class svm classification system based on methods of self-learning and error filtering
publishDate	2007
url	http://ndltd.ncl.edu.tw/handle/21380011516814819107
work_keys_str_mv	AT chihhsiunghuang amulticlasssvmclassificationsystembasedonmethodsofselflearninganderrorfiltering AT huángzhìxióng amulticlasssvmclassificationsystembasedonmethodsofselflearninganderrorfiltering AT chihhsiunghuang lìyòngzìdòngxuéxíyǔcuòwùguòlǜfāngfǎdeduōlèibiézhīyuánxiàngliàngjīfēnlèixìtǒng AT huángzhìxióng lìyòngzìdòngxuéxíyǔcuòwùguòlǜfāngfǎdeduōlèibiézhīyuánxiàngliàngjīfēnlèixìtǒng AT chihhsiunghuang multiclasssvmclassificationsystembasedonmethodsofselflearninganderrorfiltering AT huángzhìxióng multiclasssvmclassificationsystembasedonmethodsofselflearninganderrorfiltering
_version_	1716845066728767488

A Multi-Class SVM Classification System Based on Methods of Self-Learning and Error Filtering

Similar Items