Knowledge Initialisation for Support Vector Machines

Diederich, Joachim and Barakat, Nahla (2004). Knowledge Initialisation for Support Vector Machines. In: N. Kasabov and Z. S. H. Chan Conference on Neuro-Computing and Evolving Intelligence (KEDRI 2004), Auckland, New Zealand, (). 13-15 December, 2004.

Attached Files (Some files may be inaccessible until you login with your UQ eSpace credentials)
Name Description MIMEType Size Downloads
NCEI_04_Knowledg.pdf NCEI_04_Knowledg.pdf application/pdf 113.33KB 194
Author Diederich, Joachim
Barakat, Nahla
Title of paper Knowledge Initialisation for Support Vector Machines
Conference name Conference on Neuro-Computing and Evolving Intelligence (KEDRI 2004)
Conference location Auckland, New Zealand
Conference dates 13-15 December, 2004
Publication Year 2004
Sub-type Fully published paper
Editor N. Kasabov
Z. S. H. Chan
Language eng
Abstract/Summary Since their introduction more than a decade ago, support vector machines (SVMs) have shown good performance in a number of application areas, including text classification, pattern recognition and bioinformatics. However, the success of SVMs comes at a cost - there is no way to utilise prior knowledge. SVMs are purely inductive learning machines. In this paper, a novel approach for rule initialisation for support vector machines is presented. The application domain is medical diagnosis. The approach presented here uses domain knowledge in the form of propositional rules to create a virtual data set to bias an SVM. The virtual data set is combined with real data for SVM learning. Knowledge initialisation results in better classification accuracy and enhanced rule quality compared with purely inductive learning.
Subjects 280200 Artificial Intelligence and Signal and Image Processing
080107 Natural Language Processing
Keyword knowledge initialisation
machine learning
support vector machines
References [1] Fung G., Mangasarian O., Shavlik J., "Knowledge-Based Support Vector Machine Classifiers",Advances in Neural Information Processing Systems, 2003,Vol. 15, pp: 521-528. [2] Kasabov N., "On-line learning, reasoning, rule extraction and aggregation in locally optimized evolving fuzzy neural networks", Neurocomputing, 2001, 1207, pp 1-21. [3] Scholkopf, B., Simard P. Y., Smola A. J., Vapnik V. N., "Prior knowledge in support vector kernels" Advances in neural information processing systems, 1998, Vol. 10, pp 640-646. [4] Tickle, A. B., Andrews, R., Golea, M., Diederich, J., "The truth will come to light: directions and challenges in extracting the knowledge embedded within trained artificial neural network". IEEE Transactions on Neural Networks 9, 6, 1998, pp. 1057-1068.
Q-Index Code EX

Version Filter Type
Citation counts: Google Scholar Search Google Scholar
Access Statistics: 221 Abstract Views, 194 File Downloads  -  Detailed Statistics
Created: Mon, 11 Apr 2005, 10:00:00 EST by Joachim Diederich on behalf of Centre for On-Line Health