Publication Details

Increasing Classification Accuracy in LibSVM Using String Kernel Functions

HOMOLIAK, I. Increasing Classification Accuracy in LibSVM Using String Kernel Functions. Proceedings of the 18th Conference STUDENT EEICT 2012. Student EEICT, Volume 2. Brno: Faculty of Electrical Engineering and Communication BUT, 2012. p. 281-283. ISBN: 978-80-214-4461-4.
Czech title
Zvýšení úspěšnosti klasifikace v libSVM s použitím řetězcových fukcí
Type
conference paper
Language
English
Authors
Homoliak Ivan, doc. Ing., Ph.D. (DITS)
Keywords

classification, libSVM, spam detection, string kernel functions

Abstract

This publication deal with classification of text data using string kernel functions and SVM. There are discussed various experiments with string kernel functions and SVM.

Annotation

This paper explores dependencies of text classification used with string kernel functions. There are described experiments with single string kernel function and also experiments with combinations of them with arithmetic operations of addition and multiplication. Gathered results are applied to detect spam messages of e-mail communication.

Published
2012
Pages
281–283
Proceedings
Proceedings of the 18th Conference STUDENT EEICT 2012
Series
Student EEICT, Volume 2
ISBN
978-80-214-4461-4
Publisher
Faculty of Electrical Engineering and Communication BUT
Place
Brno
BibTeX 
@inproceedings{BUT98568,
  author="Ivan {Homoliak}",
  title="Increasing Classification Accuracy in LibSVM Using String Kernel Functions",
  booktitle="Proceedings of the 18th Conference STUDENT EEICT 2012",
  year="2012",
  series="Student EEICT, Volume 2",
  pages="281--283",
  publisher="Faculty of Electrical Engineering and Communication BUT",
  address="Brno",
  isbn="978-80-214-4461-4",
  url="https://www.fit.vut.cz/research/publication/10301/"
}
Files
Back to top