Classification of Small Datasets: Why Using Class-Based Weighting Measures?

Flavien Bouillot 1, 2 Pascal Poncelet 1 Mathieu Roche 3, 1
1 ADVANSE - ADVanced Analytics for data SciencE
LIRMM - Laboratoire d'Informatique de Robotique et de Microélectronique de Montpellier
2 Itesoft R&D
3 UMR TETIS - Territoires, Environnement, Télédétection et Information Spatiale
Abstract : In text classification, providing an efficient classifier even if the number of documents involved in the learning step is small remains an important issue. In this paper we evaluate the performance of traditional classification methods to better evaluate their limitation in the learning phase when dealing with small amount of documents. We thus propose a new way for weighting features which are used for classifying. These features have been integrated in two well known classifiers: Class-Feature-Centroid and Naïve Bayes, and evaluations have been performed on two real datasets. We have also investigated the influence on parameters such as number of classes, documents or words in the classification. Experiments have shown the efficiency of our proposal relatively to state of the art classification methods. Either with a very few amount of data or with a small number of features that can be extracted from poor content documents, we show that our approach performs well.
Document type :
Conference papers
Complete list of metadatas
https://hal-lirmm.ccsd.cnrs.fr/lirmm-01054900
Contributor : Mathieu Roche <>
Submitted on : Tuesday, November 6, 2018 - 1:20:55 PM
Last modification on : Friday, March 29, 2019 - 9:12:06 AM
Long-term archiving on : Thursday, February 7, 2019 - 3:15:59 PM

File

lirmm-01054900.pdf
Files produced by the author(s)

Identifiers

Collections

LIRMM | MIPS | AGROPOLIS | CIRAD | ADVANSE | IRSTEA | AGROPARISTECH | UNIV-MONTPELLIER

Citation

Flavien Bouillot, Pascal Poncelet, Mathieu Roche. Classification of Small Datasets: Why Using Class-Based Weighting Measures?. ISMIS: International Symposium on Methodologies for Intelligent Systems, Jun 2014, Roskilde, Denmark. pp.345-354, ⟨10.1007/978-3-319-08326-1_35⟩. ⟨lirmm-01054900⟩

Export

BibTeX TEI DC DCterms EndNote

Share

Metrics

Record views

103

Files downloads

51

Contact

support.ccsd.cnrs.fr
hal.support@ccsd.cnrs.fr
Logo CCSD