A Distributed Feature Extraction Method Based on MapReduce

For data with complex dimensions, Kernel Principal Component Analysis (KPCA) is a common method of feature extraction, it has a better generalization performance. However, it often face with significant difficulties in computation on large datasets. Therefore, we developed a parallel KPCA algorithm based on MapReduce (MRKPCA) in this paper, it's a parallel computing method which can run in a cluster. We apply our algorithm to an experiment of datasets of face recognition, the experimental results show that our algorithm can scale well and has better timeliness for some specific problems which require a lot of computation, and it can significantly reduce computation time.