Home // INTENSIVE 2013, The Fifth International Conference on Resource Intensive Applications and Services // View article

Design of Parallel Architectures of Classifiers Suitable for Intensive Data Processing

Authors:
Boguslaw Cyganek
Kazimierz Wiatr

Keywords: ensemble of classifiers; OC-SVM; data processing; GPU implementation; image segmentation

Abstract:
Processing of visual data, such as object recognition and image segmentation, is based on data classification. In this paper architectures of ensembles of classifiers are discussed which show superior accuracy in respect to a single classifier. To achieve comparable response time the parallel computer architectures need to be considered, however. In the paper we present a parallel implementation on a graphic card of an ensemble of one-class support vector machines for image segmentation. We show that the parallel architecture of the ensemble of classifiers allows both, the high accuracy and speed up factor of two orders of magnitude compared with the serial software implementation.

Pages: 14 to 19

Copyright: Copyright (c) IARIA, 2013

Publication date: March 24, 2013

Published in: conference

ISBN: 978-1-61208-258-5

Location: Lisbon, Portugal

Dates: from March 24, 2013 to March 29, 2013