A parallel approach for SVM with multi-core CPU

Modern multi-core CPU are predominant in improving performance of parallel applications. We describe an approach for support vector machines (SVM) that can be parallelized efficiently and scales to large problems with large training vectors. Parallel programming in .NET framework using SVM on multi-core computers are studied. Experiment results show that the serial program can only use one core of multi-core computers. Meanwhile, the parallel efficiency of the program using SVM is very high on 2-core and 4-core computers respectively.