High-performance integrated platform for visual mining of large high dimensional data sets

• Grant number: 13/22902-6
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: February 01, 2014
• End date: January 31, 2015
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computer Systems
• Principal Investigator: Almir Olivette Artero
• Grantee: Ricardo Sussumu Ywata
• Host Institution: Faculdade de Ciências e Tecnologia (FCT). Universidade Estadual Paulista (UNESP). Campus de Presidente Prudente. Presidente Prudente , SP, Brazil

Abstract

This project aims to build a high performance platform for visual data mining, comprising the main process techniques involved in visual data exploration, so as to avoid the need to use different tools, as normally occurs. Because the processing usually occurs in databases with a large number of records and higher dimensions, we will investigate how these techniques can be implemented in order to use the maximum resources of parallel programming through CUDA (Compute Unified Device Architecture). The first stage of this work is to study deeply the techniques of selection and ordering of attributes (SBAA, SBAR and Analysis of Variance) and information visualization techniques (Parallel Coordinates and Viz3D) to be implemented with the maximum parallelization of code in CUDA. Then, the techniques will be tested with data sets with large amounts of records and carried out experiments to evaluate the gains from the distribution of processing between the GPU cores.