The well known principle of curse of dimensionality links both dimensions of a dataset stating that as dimensionality increases samples become too sparse to effectively extract knowledge. Hence dimensionality reduction is essential when there are many features and not sufficient samples.We describe an algorithm for unsupervised dimensionality reduction that exploits a model of the hybridization of rough and fuzzy sets. Rough set theory and fuzzy logic are mathematical frameworks for granular computing forming a theoretical basis for the treatment of uncertainty in many real–world problems. The hybrid notion of rough fuzzy sets comes from the combination of these two models of uncertainty and helps to exploit, at the same time, properties like coarseness and vagueness. Experimental results demonstrated that the proposed approach can effectively reduce dataset dimensionality whilst retaining useful features when class labels are unknown or missing.

Feature Selection Through Composition of Rough-Fuzzy Sets

PETROSINO, Alfredo
2016-01-01

Abstract

The well known principle of curse of dimensionality links both dimensions of a dataset stating that as dimensionality increases samples become too sparse to effectively extract knowledge. Hence dimensionality reduction is essential when there are many features and not sufficient samples.We describe an algorithm for unsupervised dimensionality reduction that exploits a model of the hybridization of rough and fuzzy sets. Rough set theory and fuzzy logic are mathematical frameworks for granular computing forming a theoretical basis for the treatment of uncertainty in many real–world problems. The hybrid notion of rough fuzzy sets comes from the combination of these two models of uncertainty and helps to exploit, at the same time, properties like coarseness and vagueness. Experimental results demonstrated that the proposed approach can effectively reduce dataset dimensionality whilst retaining useful features when class labels are unknown or missing.
2016
978-3-319-52961-5
978-3-319-52962-2
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11367/59829
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 3
  • ???jsp.display-item.citation.isi??? ND
social impact