Combining LSI with other Classifiers to Improve Accuracy...
Transcript of Combining LSI with other Classifiers to Improve Accuracy...
![Page 1: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/1.jpg)
Combining LSI with other Classifiers to ImproveAccuracy of Single-label Text Categorization
Ana Cardoso-Cachopo Arlindo Oliveira
Instituto Superior Tecnico — Technical University of Lisbon / INESC-ID
EWLSATEL, March 2007
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 1 / 11
![Page 2: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/2.jpg)
Outline
1 Introduction
2 Classification Methods
3 Combinations Between Methods
4 Experimental Setup
5 Experimental Results
6 Conclusions and Future Work
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 2 / 11
![Page 3: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/3.jpg)
Outline
1 Introduction
2 Classification Methods
3 Combinations Between Methods
4 Experimental Setup
5 Experimental Results
6 Conclusions and Future Work
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 2 / 11
![Page 4: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/4.jpg)
Outline
1 Introduction
2 Classification Methods
3 Combinations Between Methods
4 Experimental Setup
5 Experimental Results
6 Conclusions and Future Work
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 2 / 11
![Page 5: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/5.jpg)
Outline
1 Introduction
2 Classification Methods
3 Combinations Between Methods
4 Experimental Setup
5 Experimental Results
6 Conclusions and Future Work
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 2 / 11
![Page 6: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/6.jpg)
Outline
1 Introduction
2 Classification Methods
3 Combinations Between Methods
4 Experimental Setup
5 Experimental Results
6 Conclusions and Future Work
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 2 / 11
![Page 7: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/7.jpg)
Outline
1 Introduction
2 Classification Methods
3 Combinations Between Methods
4 Experimental Setup
5 Experimental Results
6 Conclusions and Future Work
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 2 / 11
![Page 8: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/8.jpg)
Introduction
Text Classification
Single-label
Classification MethodsI VectorI k-NNI SVMI LSI
Goal: improve Accuracy
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 3 / 11
![Page 9: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/9.jpg)
Introduction
Text Classification
Single-label
Classification MethodsI VectorI k-NNI SVMI LSI
Goal: improve Accuracy
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 3 / 11
![Page 10: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/10.jpg)
Introduction
Text Classification
Single-label
Classification MethodsI VectorI k-NNI SVMI LSI
Goal: improve Accuracy
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 3 / 11
![Page 11: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/11.jpg)
Introduction
Text Classification
Single-label
Classification MethodsI VectorI k-NNI SVMI LSI
Goal: improve Accuracy
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 3 / 11
![Page 12: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/12.jpg)
Introduction
Text Classification
Single-label
Classification MethodsI VectorI k-NNI SVMI LSI
Goal: improve Accuracy
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 3 / 11
![Page 13: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/13.jpg)
Introduction
Text Classification
Single-label
Classification MethodsI VectorI k-NNI SVMI LSI
Goal: improve Accuracy
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 3 / 11
![Page 14: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/14.jpg)
Introduction
Text Classification
Single-label
Classification MethodsI VectorI k-NNI SVMI LSI
Goal: improve Accuracy
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 3 / 11
![Page 15: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/15.jpg)
Introduction
Text Classification
Single-label
Classification MethodsI VectorI k-NNI SVMI LSI
Goal: improve Accuracy
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 3 / 11
![Page 16: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/16.jpg)
Classification Methods
p dimensionalterm space
s << pdimensional
concept space
SVM
k-NN
Vector
LSI
Cosine similarity
k-NN + Cosine similarity
Kernel Voting strategy
SVD Cosine similarity
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 4 / 11
![Page 17: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/17.jpg)
Combinations Between Methods
p dimensionalterm space
s << pdimensional
concept space
k-NN-LSI
SVM-LSI
k-NN + Cosine similarity
Kernel + Voting strategy
SVD
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 5 / 11
![Page 18: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/18.jpg)
Experimental Setup
Methods (6 already mentioned + Dumb)
DatasetsI Bank´s Data - Bank37I Reuters 21578 - R8, R52I 20 Newsgroups - 20NgI Web Knowledge Base - Web4I Cade - Cade12
Evaluation Measure
Accuracy =#Correctly classified documents
#Total documents
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 6 / 11
![Page 19: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/19.jpg)
Experimental Setup
Methods (6 already mentioned + Dumb)
DatasetsI Bank´s Data - Bank37I Reuters 21578 - R8, R52I 20 Newsgroups - 20NgI Web Knowledge Base - Web4I Cade - Cade12
Evaluation Measure
Accuracy =#Correctly classified documents
#Total documents
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 6 / 11
![Page 20: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/20.jpg)
Experimental Setup
Methods (6 already mentioned + Dumb)
DatasetsI Bank´s Data - Bank37I Reuters 21578 - R8, R52I 20 Newsgroups - 20NgI Web Knowledge Base - Web4I Cade - Cade12
Evaluation Measure
Accuracy =#Correctly classified documents
#Total documents
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 6 / 11
![Page 21: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/21.jpg)
Experimental Setup
Methods (6 already mentioned + Dumb)
DatasetsI Bank´s Data - Bank37I Reuters 21578 - R8, R52I 20 Newsgroups - 20NgI Web Knowledge Base - Web4I Cade - Cade12
Evaluation Measure
Accuracy =#Correctly classified documents
#Total documents
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 6 / 11
![Page 22: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/22.jpg)
Experimental Setup
Methods (6 already mentioned + Dumb)
DatasetsI Bank´s Data - Bank37I Reuters 21578 - R8, R52I 20 Newsgroups - 20NgI Web Knowledge Base - Web4I Cade - Cade12
Evaluation Measure
Accuracy =#Correctly classified documents
#Total documents
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 6 / 11
![Page 23: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/23.jpg)
Experimental Setup
Methods (6 already mentioned + Dumb)
DatasetsI Bank´s Data - Bank37I Reuters 21578 - R8, R52I 20 Newsgroups - 20NgI Web Knowledge Base - Web4I Cade - Cade12
Evaluation Measure
Accuracy =#Correctly classified documents
#Total documents
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 6 / 11
![Page 24: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/24.jpg)
Experimental Setup
Methods (6 already mentioned + Dumb)
DatasetsI Bank´s Data - Bank37I Reuters 21578 - R8, R52I 20 Newsgroups - 20NgI Web Knowledge Base - Web4I Cade - Cade12
Evaluation Measure
Accuracy =#Correctly classified documents
#Total documents
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 6 / 11
![Page 25: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/25.jpg)
Experimental Setup
Methods (6 already mentioned + Dumb)
DatasetsI Bank´s Data - Bank37I Reuters 21578 - R8, R52I 20 Newsgroups - 20NgI Web Knowledge Base - Web4I Cade - Cade12
Evaluation Measure
Accuracy =#Correctly classified documents
#Total documents
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 6 / 11
![Page 26: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/26.jpg)
Characteristics of the Datasets
Train Test Total Smallest LargestDocs Docs Docs Class Class
Bank37 928 463 1391 5 346
20Ng 11293 7528 18821 628 999
R8 5485 2189 7674 51 3923
R52 6532 2568 9100 3 3923
Web4 2803 1396 4199 504 1641
Cade12 27322 13661 40983 625 8473
Numbers of documents for the datasets: number of training documents,number of test documents, total number of documents, number ofdocuments in the smallest class, and number of documents in the largestclass.
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 7 / 11
![Page 27: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/27.jpg)
Characteristics of the Datasets
Train Test Total Smallest LargestDocs Docs Docs Class Class
Bank37 928 463 1391 5 346
20Ng 11293 7528 18821 628 999
R8 5485 2189 7674 51 3923
R52 6532 2568 9100 3 3923
Web4 2803 1396 4199 504 1641
Cade12 27322 13661 40983 625 8473
Numbers of documents for the datasets: number of training documents,number of test documents, total number of documents, number ofdocuments in the smallest class, and number of documents in the largestclass.
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 7 / 11
![Page 28: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/28.jpg)
Characteristics of the Datasets
Train Test Total Smallest LargestDocs Docs Docs Class Class
Bank37 928 463 1391 5 346
20Ng 11293 7528 18821 628 999
R8 5485 2189 7674 51 3923
R52 6532 2568 9100 3 3923
Web4 2803 1396 4199 504 1641
Cade12 27322 13661 40983 625 8473
Numbers of documents for the datasets: number of training documents,number of test documents, total number of documents, number ofdocuments in the smallest class, and number of documents in the largestclass.
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 7 / 11
![Page 29: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/29.jpg)
Characteristics of the Datasets
Train Test Total Smallest LargestDocs Docs Docs Class Class
Bank37 928 463 1391 5 346
20Ng 11293 7528 18821 628 999
R8 5485 2189 7674 51 3923
R52 6532 2568 9100 3 3923
Web4 2803 1396 4199 504 1641
Cade12 27322 13661 40983 625 8473
Numbers of documents for the datasets: number of training documents,number of test documents, total number of documents, number ofdocuments in the smallest class, and number of documents in the largestclass.
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 7 / 11
![Page 30: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/30.jpg)
Experimental Results
0.0
0.2
0.4
0.6
0.8
1.0Dumb
Vector
k-NN
SVM
LSI
k-NN-LSI
SVM-LSI
Bank37 20Ng R8 R52 Web4 Cade12
Accuracy values for the six datasets using each method.
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 8 / 11
![Page 31: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/31.jpg)
Experimental Results
0.0
0.2
0.4
0.6
0.8
1.0Dumb
Vector
k-NN
SVM
LSI
k-NN-LSI
SVM-LSI
Bank37 20Ng R8 R52 Web4 Cade12
Accuracy values for the six datasets using each method.
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 8 / 11
![Page 32: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/32.jpg)
Experimental Results
0.0
0.2
0.4
0.6
0.8
1.0Dumb
Vector
k-NN
SVM
LSI
k-NN-LSI
SVM-LSI
Bank37 20Ng R8 R52 Web4 Cade12
Accuracy values for the six datasets using each method.
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 8 / 11
![Page 33: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/33.jpg)
Experimental Results
0.0
0.2
0.4
0.6
0.8
1.0Dumb
Vector
k-NN
SVM
LSI
k-NN-LSI
SVM-LSI
Bank37 20Ng R8 R52 Web4 Cade12
Accuracy values for the six datasets using each method.
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 8 / 11
![Page 34: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/34.jpg)
Experimental Results
0.0
0.2
0.4
0.6
0.8
1.0Dumb
Vector
k-NN
SVM
LSI
k-NN-LSI
SVM-LSI
Bank37 20Ng R8 R52 Web4 Cade12
Accuracy values for the six datasets using each method.
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 8 / 11
![Page 35: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/35.jpg)
Experimental Results
0.0
0.2
0.4
0.6
0.8
1.0Dumb
Vector
k-NN
SVM
LSI
k-NN-LSI
SVM-LSI
Bank37 20Ng R8 R52 Web4 Cade12
Accuracy values for the six datasets using each method.
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 8 / 11
![Page 36: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/36.jpg)
Experimental Results
Dataset Dumb Vector k-NN SVM LSIk-NNLSI
SVMLSI
Bank37 0.2505 0.8359 0.8423 0.9071 0.8531 0.8488 0.917920Ng 0.0530 0.7240 0.7593 0.8284 0.7491 0.7557 0.7775R8 0.4947 0.7889 0.8524 0.9698 0.9411 0.9488 0.9680R52 0.4217 0.7687 0.8322 0.9377 0.9093 0.9100 0.9311Web4 0.3897 0.6447 0.7256 0.8582 0.7357 0.7908 0.8897Cade12 0.2083 0.4142 0.5120 0.5284 0.4329 0.4880 0.5465
Average 0.3030 0.6961 0.7540 0.8383 0.7702 0.7904 0.8385
Accuracy values for the six datasets using each method, and averageAccuracy for each method over all the datasets.
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 9 / 11
![Page 37: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/37.jpg)
Experimental Results
Dataset Dumb Vector k-NN SVM LSIk-NNLSI
SVMLSI
Bank37 0.2505 0.8359 0.8423 0.9071 0.8531 0.8488 0.917920Ng 0.0530 0.7240 0.7593 0.8284 0.7491 0.7557 0.7775R8 0.4947 0.7889 0.8524 0.9698 0.9411 0.9488 0.9680R52 0.4217 0.7687 0.8322 0.9377 0.9093 0.9100 0.9311Web4 0.3897 0.6447 0.7256 0.8582 0.7357 0.7908 0.8897Cade12 0.2083 0.4142 0.5120 0.5284 0.4329 0.4880 0.5465
Average 0.3030 0.6961 0.7540 0.8383 0.7702 0.7904 0.8385
Accuracy values for the six datasets using each method, and averageAccuracy for each method over all the datasets.
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 9 / 11
![Page 38: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/38.jpg)
Experimental Results
Dataset Dumb Vector k-NN SVM LSIk-NNLSI
SVMLSI
Bank37 0.2505 0.8359 0.8423 0.9071 0.8531 0.8488 0.917920Ng 0.0530 0.7240 0.7593 0.8284 0.7491 0.7557 0.7775R8 0.4947 0.7889 0.8524 0.9698 0.9411 0.9488 0.9680R52 0.4217 0.7687 0.8322 0.9377 0.9093 0.9100 0.9311Web4 0.3897 0.6447 0.7256 0.8582 0.7357 0.7908 0.8897Cade12 0.2083 0.4142 0.5120 0.5284 0.4329 0.4880 0.5465
Average 0.3030 0.6961 0.7540 0.8383 0.7702 0.7904 0.8385
Accuracy values for the six datasets using each method, and averageAccuracy for each method over all the datasets.
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 9 / 11
![Page 39: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/39.jpg)
Experimental Results
Dataset Dumb Vector k-NN SVM LSIk-NNLSI
SVMLSI
Bank37 0.2505 0.8359 0.8423 0.9071 0.8531 0.8488 0.917920Ng 0.0530 0.7240 0.7593 0.8284 0.7491 0.7557 0.7775R8 0.4947 0.7889 0.8524 0.9698 0.9411 0.9488 0.9680R52 0.4217 0.7687 0.8322 0.9377 0.9093 0.9100 0.9311Web4 0.3897 0.6447 0.7256 0.8582 0.7357 0.7908 0.8897Cade12 0.2083 0.4142 0.5120 0.5284 0.4329 0.4880 0.5465
Average 0.3030 0.6961 0.7540 0.8383 0.7702 0.7904 0.8385
Accuracy values for the six datasets using each method, and averageAccuracy for each method over all the datasets.
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 9 / 11
![Page 40: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/40.jpg)
Experimental Results
Dataset Dumb Vector k-NN SVM LSIk-NNLSI
SVMLSI
Bank37 0.2505 0.8359 0.8423 0.9071 0.8531 0.8488 0.917920Ng 0.0530 0.7240 0.7593 0.8284 0.7491 0.7557 0.7775R8 0.4947 0.7889 0.8524 0.9698 0.9411 0.9488 0.9680R52 0.4217 0.7687 0.8322 0.9377 0.9093 0.9100 0.9311Web4 0.3897 0.6447 0.7256 0.8582 0.7357 0.7908 0.8897Cade12 0.2083 0.4142 0.5120 0.5284 0.4329 0.4880 0.5465
Average 0.3030 0.6961 0.7540 0.8383 0.7702 0.7904 0.8385
Accuracy values for the six datasets using each method, and averageAccuracy for each method over all the datasets.
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 9 / 11
![Page 41: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/41.jpg)
Conclusions and Future Work
Very good Accuracy for some datasets.
It is worth pursuing this line of research by testing more combinationsbetween the method´s parameters.
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 10 / 11
![Page 42: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/42.jpg)
Conclusions and Future Work
Very good Accuracy for some datasets.
It is worth pursuing this line of research by testing more combinationsbetween the method´s parameters.
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 10 / 11
![Page 43: Combining LSI with other Classifiers to Improve Accuracy ...web.ist.utl.pt/~acardoso/docs/...presentation.pdf · Combining LSI with other Classifiers to Improve Accuracy of Single-label](https://reader034.fdocuments.in/reader034/viewer/2022051922/600fb7eb2a33364e735f0d4d/html5/thumbnails/43.jpg)
Thank You.
Any Questions?
(IST-TULisbon/INESC-ID) Ana Cardoso-Cachopo EWLSATEL, March 2007 11 / 11