逻辑回归参数训练过程的MapReduce并行化实现

The Parallelism Implementation of Logistic Regression Parameter Training on Map Reduce

在逻辑回归算法的实际工程应用中,传统的单机参数训练方法在处理大规模的数据时效率低下.采用并行计算框架实现对训练数据的并行处理,能够有效解决这一问题.基于MapReduce分布式并行计算框架,提出逻辑回归参数训练过程在MapReduce集群上的实现方法并用实验加以验证.

In the engineering applications of Logistic Regression with large input data, the single computation node was inefficient. It could can be addressed by parallelism computation mode. It proposed an implementation of the training process on MapReduce framework and tested the effectiveness of this method with an experiment pin the paper.