Modified fast PCA algorithm on GPU architecture

Abstract

Recognition task is a hard problem due to the high dimension of input image data. The principal component analysis (PCA) is the one of the most popular algorithms for reducing the dimensionality. The main constraint of PCA is the execution time in terms of updating when new data is included; therefore, parallel computation is needed. Opening the GPU architectures to general purpose computation allows performing parallel computation on a powerful platform. In this paper the modified version of fast PCA (MFPCA) algorithm is presented on the GPU architecture and also the suitability of the algorithm for face recognition task is discussed. The performance and efficiency of MFPCA algorithm is studied on large-scale datasets. Experimental results show a decrease of the MFPCA algorithm execution time while preserving the quality of the results.