Clustering and swarm intelligence with parallel computing using GPU

• Grant number: 13/23027-1
• Support Opportunities: Regular Research Grants
• Start date: May 01, 2014
• End date: April 30, 2015
• Field of knowledge: Physical Sciences and Mathematics - Computer Science
• Principal Investigator: Adriane Beatriz de Souza Serapião
• Grantee: Adriane Beatriz de Souza Serapião
• Host Institution: Instituto de Geociências e Ciências Exatas (IGCE). Universidade Estadual Paulista (UNESP). Campus de Rio Claro. Rio Claro , SP, Brazil
• Associated researchers: Veronica Oliveira de Carvalho

Abstract

In recent years, new technologies provided a huge increase in storage capacity and data processing, creating the demand for analysis of large data volumes captured by scientific instruments or generated by simulations, and resulting in data with a high dimensionality. However, the computational tools for inspection and knowledge extraction from databases not developed in the same level. The cluster analysis (clustering) plays a central role in Data Mining and Knowledge Discovery in databases, assisting in the resolution of the data growing problem, and producing a data separation model to discover groups of similar objects. In cluster analysis, each group, called cluster, consists of objects that are similar among them and different of the objects of other groups. A similarity measure based on a distance metric is used to define the proximity between a pair of objects. The organization of data in a cluster is performed according to a given similarity by using an unsupervised learning approach, with a data set not labeled, from which it seeks to find out how objects are arranged. In the partitional clustering approach the algorithms seek to determine the centers of the clusters and the number of cluster according to a criterion in order to produce the best separation between the data. Thus, the partitional clustering can be viewed as an optimization task. Swarm Intelligence algorithms are quite promising for executing this kind of task, since metaheuristic methods are widely used in optimization problems. This project aims to develop new methods of clustering using Swarm Intelligence algorithms. Due to the inherently parallel nature of these algorithms, the proposed clustering methods will be implemented in graphics processing units (GPU). (AU)

Scientific publications

(References retrieved automatically from Web of Science and SciELO through information on FAPESP grants and their corresponding numbers as mentioned in the publications by the authors)

SERAPIAO, ADRIANE B. S.; CORREA, GUILHERME S.; GONCALVES, FELIPE B.; CARVALHO, VERONICA O.. Combining K-Means and K-Harmonic with Fish School Search Algorithm for data clustering task on graphics processing units. APPLIED SOFT COMPUTING, v. 41, p. 290-304, APR 2016. (13/23027-1, 13/08730-8, 13/08741-0)