ELM-MapReduce: MapReduce accelerated extreme learning machine for big spatial data analysis

Land cover classification of remote sensing (RS) data plays a key role in various spatio-temporal applications. Moreover, scalability and efficiency have become the most important challenges because of increasing RS data. In this paper, we propose a novel MapReduce accelerated extreme learning machine (ELM) ensemble classifier called ELM-MapReduce for large scale land cover classification. First, ELM-MapReduce adopts ELM ensemble learning algorithm with higher accuracy and stability. Second, ELM-MapReduce is accelerated by MapReduce for higher scalability and efficiency. Third, the experiments on large scale real world RS data have proven the advantages of ELM-MapReduce.