Journal of Engineering and Applied Sciences
ISSN: Online 1818-7803ISSN: Print 1816-949x
Abstract
One of the drawbacks of MapReduce characteristic is overlap communication. It causes implementation inefficiency in the GPUs environment. However, this can be overcome using incremental reduction method. This method will enhance the communication process on GPUs environment as an alternative to execution using CPU. This enhancement is based on Python with support of CUDA technologies which can execute this whole process in GPUs environment. In order to achieve the good performance, this study is proposing to design the MapReduce with incremental reduction and then to construct it and finally to test the enhancement method to the self-organizing map with handwriting dataset.
How to cite this article:
Faeez Abd Rashid, Noor Elaiza Abd Khalid, Muhammad Firdaus Mustapha and Mazani Manaf. Parallel Self-Organizing Map using MapReduce in GPUs Environment.
DOI: https://doi.org/10.36478/jeasci.2018.314.320
URL: https://www.makhillpublications.co/view-article/1816-949x/jeasci.2018.314.320