Feature selection for multi-label learning

• Grant number: 12/23906-2
• Support Opportunities: Scholarships abroad - Research Internship - Doctorate
• Start date: April 01, 2013
• End date: August 31, 2013
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
• Principal Investigator: Maria Carolina Monard
• Grantee: Newton Spolaôr
• Supervisor: Grigorios D. Tsoumakas
• Host Institution: Instituto de Ciências Matemáticas e de Computação (ICMC). Universidade de São Paulo (USP). São Carlos , SP, Brazil
• Institution abroad: Aristotle University of Thessaloniki (AUTh), Greece
• Associated to the scholarship: 11/02393-4 - Feature Selection for Multi-label Learning, BP.DR

Abstract

Feature Selection (FS) is an important task in machine learning, which caneffectively reduce the dataset dimensionality by removing irrelevantand/or redundant features. Although a large body of research dealswith FS in single-label data, in which several measures have been proposedto filter out irrelevant features, this is not the case for multi-labeldata where each instance is labeled with a set of labels. In thisscenario, the additional problem of label dependence should be consideredby the feature importance measures. This work proposes to investigatefeature importance measures which take into account label dependence,aiming to find new FS methods for multi-label data. (AU)