A Fast Parallel SVM Algorithm for Massive Classification Tasks

The new parallel incremental Support VectorMachine (SVM) algorithm aims at classifying very large datasets on graphics processing units (GPUs). SVM and kernel related methods have shown to build accurate models but the learning task usually needs a quadratic programming, so that the learning task for large datasets requires big memory capacity and a long time. We extend the recent finite Newton classifier for building a parallel incremental algorithm. The new algorithm uses graphics processors to gain high performance at low cost. Numerical test results on UCI, Delve dataset repositories showed that our parallel incremental algorithm using GPUs is about 45 times faster than a CPU implementation and often significantly over 100 times faster than state-of-the-art algorithms LibSVM, SVM-perf and CB-SVM.