VISUAL SVM
François Poulet
2005
Abstract
We present a cooperative approach using both Support Vector Machine (SVM) algorithms and visualization methods. SVM are widely used today and often give high quality results, but they are used as "black-box", (it is very difficult to explain the obtained results) and cannot treat easily very large datasets. We have developed graphical methods to help the user to evaluate and explain the SVM results. The first method is a graphical representation of the separating frontier quality (it is presented for the SVM case, but can be used for any other boundary like decision tree cuts, regression lines, etc). Then it is linked with other graphical methods to help the user explaining SVM results. The information provided by these graphical methods can also be used in the SVM parameter tuning stage. These graphical methods are then used together with automatic algorithms to deal with very large datasets on standard personal computers. We present an evaluation of our approach with the UCI and the Kent Ridge Bio-medical data sets.
DownloadPaper Citation
in Harvard Style
Poulet F. (2005). VISUAL SVM . In Proceedings of the Seventh International Conference on Enterprise Information Systems - Volume 2: ICEIS, ISBN 972-8865-19-8, pages 309-314. DOI: 10.5220/0002521003090314
in Bibtex Style
@conference{iceis05,
author={François Poulet},
title={VISUAL SVM},
booktitle={Proceedings of the Seventh International Conference on Enterprise Information Systems - Volume 2: ICEIS,},
year={2005},
pages={309-314},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0002521003090314},
isbn={972-8865-19-8},
}
in EndNote Style
TY - CONF
JO - Proceedings of the Seventh International Conference on Enterprise Information Systems - Volume 2: ICEIS,
TI - VISUAL SVM
SN - 972-8865-19-8
AU - Poulet F.
PY - 2005
SP - 309
EP - 314
DO - 10.5220/0002521003090314