Advanced search
1 file | 164.31 KB Add to list

Multi threshold FRPS: a new approach to fuzzy rough set prototype selection

Nele Verbiest (UGent)
Author
Organization
Abstract
Prototype Selection (PS) is the preprocessing technique for K nearest neighbor classification that selects a subset of instances before classification takes place. The most accurate state-of-the-art PS method is Fuzzy Rough Prototype Selection (FRPS), which assesses the quality of the instances by means of the fuzzy rough positive region and automatically selects a good threshold to decide if instances should be retained in the prototype subset. In this paper we introduce a new PS method based on FRPS, called Multi Threshold FRPS (MT-FRPS). Instead of determining one threshold against which the quality of every instance is compared, we consider one threshold for each class. We evaluate MT-FRPS on 40 standard classification datasets and compare it against MT-FRPS and the state-of-the-art PS methods and show that MT-FRPS improves the accuracy of the state-of-the-art PS methods.
Keywords
fuzzy rough set theory, prototype selection, classification, CLASSIFICATION, ALGORITHM, RULE

Downloads

  • (...).pdf
    • full text
    • |
    • UGent only
    • |
    • PDF
    • |
    • 164.31 KB

Citation

Please use this url to cite or link to this publication:

MLA
Verbiest, Nele. “Multi Threshold FRPS: A New Approach to Fuzzy Rough Set Prototype Selection.” Lecture Notes in Computer Science, edited by Chris Cornelis et al., vol. 8536, Springer, 2014, pp. 83–91, doi:10.1007/978-3-319-08644-6_8.
APA
Verbiest, N. (2014). Multi threshold FRPS: a new approach to fuzzy rough set prototype selection. In C. Cornelis, M. Kryszkiewicz, D. Ślęzak, E. M. Ruiz, R. Bello, & L. Shang (Eds.), Lecture Notes in Computer Science (Vol. 8536, pp. 83–91). https://doi.org/10.1007/978-3-319-08644-6_8
Chicago author-date
Verbiest, Nele. 2014. “Multi Threshold FRPS: A New Approach to Fuzzy Rough Set Prototype Selection.” In Lecture Notes in Computer Science, edited by Chris Cornelis, Marzena Kryszkiewicz, Dominik Ślęzak, Ernestina Menasalvas Ruiz, Rafael Bello, and Lin Shang, 8536:83–91. Berlin, Germany: Springer. https://doi.org/10.1007/978-3-319-08644-6_8.
Chicago author-date (all authors)
Verbiest, Nele. 2014. “Multi Threshold FRPS: A New Approach to Fuzzy Rough Set Prototype Selection.” In Lecture Notes in Computer Science, ed by. Chris Cornelis, Marzena Kryszkiewicz, Dominik Ślęzak, Ernestina Menasalvas Ruiz, Rafael Bello, and Lin Shang, 8536:83–91. Berlin, Germany: Springer. doi:10.1007/978-3-319-08644-6_8.
Vancouver
1.
Verbiest N. Multi threshold FRPS: a new approach to fuzzy rough set prototype selection. In: Cornelis C, Kryszkiewicz M, Ślęzak D, Ruiz EM, Bello R, Shang L, editors. Lecture Notes in Computer Science. Berlin, Germany: Springer; 2014. p. 83–91.
IEEE
[1]
N. Verbiest, “Multi threshold FRPS: a new approach to fuzzy rough set prototype selection,” in Lecture Notes in Computer Science, Granada ; Madrid, Spain, 2014, vol. 8536, pp. 83–91.
@inproceedings{5671986,
  abstract     = {{Prototype Selection (PS) is the preprocessing technique for K nearest neighbor classification that selects a subset of instances before classification takes place. The most accurate state-of-the-art PS method is Fuzzy Rough Prototype Selection (FRPS), which assesses the quality of the instances by means of the fuzzy rough positive region and automatically selects a good threshold to decide if instances should be retained in the prototype subset. In this paper we introduce a new PS method based on FRPS, called Multi Threshold FRPS (MT-FRPS). Instead of determining one threshold against which the quality of every instance is compared, we consider one threshold for each class. 
We evaluate MT-FRPS on 40 standard classification datasets and compare it against MT-FRPS and the state-of-the-art PS methods and show that MT-FRPS improves the accuracy of the state-of-the-art PS methods.}},
  author       = {{Verbiest, Nele}},
  booktitle    = {{Lecture Notes in Computer Science}},
  editor       = {{Cornelis, Chris and Kryszkiewicz, Marzena and Ślęzak, Dominik and Ruiz, Ernestina Menasalvas and Bello, Rafael and Shang, Lin}},
  isbn         = {{9783319086439}},
  issn         = {{0302-9743}},
  keywords     = {{fuzzy rough set theory,prototype selection,classification,CLASSIFICATION,ALGORITHM,RULE}},
  language     = {{eng}},
  location     = {{Granada ; Madrid, Spain}},
  pages        = {{83--91}},
  publisher    = {{Springer}},
  title        = {{Multi threshold FRPS: a new approach to fuzzy rough set prototype selection}},
  url          = {{http://doi.org/10.1007/978-3-319-08644-6_8}},
  volume       = {{8536}},
  year         = {{2014}},
}

Altmetric
View in Altmetric
Web of Science
Times cited: