Multicore Learning Algorithm

The focus of our term project is to apply the map-reduce principle to a variety of machine learning algorithms that are computationally expensive. Instead of using expensive computer clusters, we focus on implementing the framework on multi-core computer environment. On top of that, in order to apply the framework to a variety of modern machine learning algorithms, we focus on parallelize the summation operation in the algorithms by developing customized mappers and reducers. In the following sections, we will show the simplified architecture in Section 2, the implemented algorithms and the formula where the framework is applied in Section 3, and finally the experiments and its result disccision in Section 4.