PARALLELIZING GENETIC ALGORITHM ON MAP REDUCE ARCHITECTURE

Abstract

Data-intensive computation has evolved as a key component of processing large volumes of data taking advantage of massive parallelism. This computing frameworks have showcased how processing of terabytes to petabytes of raw data can be attained routinely with ease. However, Not much efforts have been done on exploring effects of evolutionary algorithms on data intensive processes. Evolutionary Algorithms like can be merged with other algorithms(presently in use) to produce better and more accurate results . Here we have presented a detailed step-bystep explanation of evolutionary computation algorithms and how they can be translated into the Map Reduce Architecture. Results have show that (1) Hadoop Architecture along with evolutionary algorithms yields better results than other contemporary algorithm in use especially in case when problem is very large and hence proves to be an excellent choice, and (2) thanks to inherent parallel processing feature of evolutionary algorithms due to which transparent linear speedups are attainable without changing the underlying data-intensive flow.