Multiple classifier systems through random selection of feature subsets

• Grant number: 12/11738-8
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: September 01, 2012
• End date: March 31, 2013
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
• Principal Investigator: Moacir Antonelli Ponti
• Grantee: Roberto Martins de Freitas
• Host Institution: Instituto de Ciências Matemáticas e de Computação (ICMC). Universidade de São Paulo (USP). São Carlos , SP, Brazil

Abstract

Classifier combination methods are effective, provided some conditions, for many pattern recognition applications. The ensemble creation method based on random feature selection can help when there are too many features available and often improves the system's accuracy. In this project we will investigate this kind of method that uses input feature manipulation, studying ways to select the features such as uniform random, weighted selection, etc. The main contribution will be the use of fewer features per trained classifier and probably an increase in the computational performance.(AU)