Author/Authors :
AsilianBidgoli, A Department of Computer Engineering - University of Kashan - Kashan, Iran , Ebrahimpour-Komleh, H Department of Computer Engineering - University of Kashan - Kashan, Iran , Askari, M Department of Computer Engineering - University of Kashan - Kashan, Iran , Mousavirad, SJ Department of Computer Engineering - University of Kashan - Kashan, Iran
Abstract :
This paper parallelizes the spatial pyramid match kernel (SPK) implementation. In the recent years, SPK has
been one of the most usable kernel methods along with support vector machine classifier with high accuracy
in object recognition. The MATLAB parallel computing toolbox is used to parallelize SPK. In this
implementation, the MATLAB Message Passing Interface (MPI) functions and the features included in the
toolbox help us obtain a good performance by the two schemes task-parallelization and data-parallelization
models. The parallel SPK algorithm runs over a cluster of computers and achieves less run time. The
observed speed-up depends on the number of CPUs and their cores. A speed-up value equal to 13 was
obtained for a configuration with up to 5 Quad processors
Keywords :
Support Vector Machine Classifier , Cluster of Computers , Parallel Computing , Spatial Pyramid Match Kernel , Object Recognition
