TY - GEN

T1 - Nearest neighbour-based fuzzy-rough feature selection

AU - Jensen, Richard

AU - MacParthaláin, Neil Seosamh

N1 - Jensen, R., MacParthaláin, N. S. (2014). Nearest neighbour-based fuzzy-rough feature selection. In C. Cornelis, M. Kryszkiewicz, & D. Slezak (Eds.), Rough Sets and Current Trends in Soft Computing: Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). (Vol. 8536 LNAI, pp. 35-46). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 8536 LNAI). Springer-Verlag

PY - 2014/1/1

Y1 - 2014/1/1

N2 - Research in the area of fuzzy-rough set theory and its application to various areas of learning have generated great interest in recent years. In particular, there has been much work in the area of feature or attribute selection. Indeed, as the number of dimensions increases, the number of data objects required in order to generate accurate models increases exponentially. Thus, feature selection (FS) has become an increasingly necessary step in model learning. The use of fuzzy-rough sets as dataset pre-processors offers much in the way of flexibility, however the underlying complexity of the subset evaluation metric often presents a problem and can result in a great deal of potentially unnecessary computational effort. This paper proposes two different novel ways to address this problem using a neighbourhood approximation step in order to alleviate the processing overhead and reduce the complexity of the evaluation metric. The experimental evaluation demonstrates that much computational effort can be avoided, and as a result the efficiency of the FS process can be improved considerably.

AB - Research in the area of fuzzy-rough set theory and its application to various areas of learning have generated great interest in recent years. In particular, there has been much work in the area of feature or attribute selection. Indeed, as the number of dimensions increases, the number of data objects required in order to generate accurate models increases exponentially. Thus, feature selection (FS) has become an increasingly necessary step in model learning. The use of fuzzy-rough sets as dataset pre-processors offers much in the way of flexibility, however the underlying complexity of the subset evaluation metric often presents a problem and can result in a great deal of potentially unnecessary computational effort. This paper proposes two different novel ways to address this problem using a neighbourhood approximation step in order to alleviate the processing overhead and reduce the complexity of the evaluation metric. The experimental evaluation demonstrates that much computational effort can be avoided, and as a result the efficiency of the FS process can be improved considerably.

KW - feature selection

KW - fuzzy-rough sets

KW - nearest neighbours

UR - http://www.scopus.com/inward/record.url?scp=84903781216&partnerID=8YFLogxK

UR - http://hdl.handle.net/2160/14227

U2 - 10.1007/978-3-319-08644-6_4

DO - 10.1007/978-3-319-08644-6_4

M3 - Conference Proceeding (Non-Journal item)

AN - SCOPUS:84903781216

SN - 9783319086439

VL - 8536 LNAI

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 35

EP - 46

BT - Rough Sets and Current Trends in Soft Computing

A2 - Cornelis, Chris

A2 - Kryszkiewicz, Marzena

A2 - Slezak, Dominik

A2 - Menasalvas Ruiz, Ernestina

A2 - Bello, Rafael

A2 - Shang, Lin

PB - Springer Nature

ER -