Techniques for unbalanced data in hierarchical classification

• Grant number: 13/15856-8
• Support Opportunities: Scholarships in Brazil - Master
• Start date: May 01, 2014
• End date: February 28, 2015
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
• Principal Investigator: André Carlos Ponce de Leon Ferreira de Carvalho
• Grantee: Victor Hugo Barella
• Host Institution: Instituto de Ciências Matemáticas e de Computação (ICMC). Universidade de São Paulo (USP). São Carlos , SP, Brazil

Abstract

Many key machine learning algorithms can not perform well for classification in scenarios in which there is disproportion between the quantities of examples from different classes. This problem is known as unbalanced data (or imbalanced classes), which is the subject of this project. Among the challenges of working with such databases is dealing with distinct distributions between groups examples and data sets in which classes are underrepresented, such as those with a small number of examples and overlap regions. Several applications have unbalanced problems, however this work aims to study such distributions in hierarchical classification problems. Like most techniques for unbalanced data are binary, it is proposed to decompose the hierarchical problem into binary subproblems. (AU)

Academic Publications

(References retrieved automatically from State of São Paulo Research Institutions)

BARELLA, Victor Hugo. Techniques for the problem of imbalanced data in hierarchical classification. 2015. Master's Dissertation - Universidade de São Paulo (USP). Instituto de Ciências Matemáticas e de Computação (ICMC/SB) São Carlos.