Number of the records: 1

Fast Dependency-Aware Feature Selection in Very-High-Dimensional Pattern Recognition Problems

SYSNO ASEP	0357265
Document Type	V - Research Report
R&D Document Type	The record was not marked in the RIV
Title	Fast Dependency-Aware Feature Selection in Very-High-Dimensional Pattern Recognition Problems
Author(s)	Somol, Petr (UTIA-B)_RID Grim, Jiří (UTIA-B)_{RID, ORCID}
Issue data	Praha: ÚTIA AV ČR, v.v.i, 2011
Series	Research Report
Series number	2295
Number of pages	9 s.
Language	eng - English
Country	CZ - Czech Republic
Keywords	feature selection, ; high dimensionality ; ranking ; generalization ; over-fitting ; stability ; classification ; pattern recognition ; machine learning
Subject RIV	BD - Theory of Information
R&D Projects	1M0572 GA MŠMT - Ministry of Education, Youth and Sports (MEYS)
	2C06019 GA MŠMT - Ministry of Education, Youth and Sports (MEYS)
CEZ	AV0Z10750506 - UTIA-B (2005-2011)
Annotation	The paper addresses the problem of making dependency-aware feature selection feasible in pattern recognition problems of very high dimensionality. The idea of individually best ranking is generalized to evaluate the contextual quality of each feature in a series of randomly generated feature subsets. Each random subset is evaluated by a criterion function of arbitrary choice (permitting functions of high complexity). Eventually, the novel dependency-aware feature rank is computed, expressing the average benefit of including a feature into feature subsets. The method is efficient and generalizes well especially in very-high-dimensional problems, where traditional context-aware feature selection methods fail due to prohibitive computational complexity or to over-fitting. The method is shown well capable of over-performing the commonly applied individual ranking which ignores important contextual information contained in data.
Workplace	Institute of Information Theory and Automation
Contact	Markéta Votavová, votavova@utia.cas.cz, Tel.: 266 052 201.
Year of Publishing	2012

Number of the records: 1