A parallel software to train linear and nonlinear SVMs for classification problems is presented, which is suitable for distributed memory multiprocessor systems. It solves the SVM quadratic programming problem by an iterative decomposition technique based on a well parallelizable gradient-projection solver for the inner subproblems. The numerical experience show the significant speedup that the proposed parallel package can achieve in training large-scale SVMs. The package is available at http://dm.unife.it/gpdt.
Parallel training of Large-Scale Kernel Machines
ZANGHIRATI, Gaetano
2006
Abstract
A parallel software to train linear and nonlinear SVMs for classification problems is presented, which is suitable for distributed memory multiprocessor systems. It solves the SVM quadratic programming problem by an iterative decomposition technique based on a well parallelizable gradient-projection solver for the inner subproblems. The numerical experience show the significant speedup that the proposed parallel package can achieve in training large-scale SVMs. The package is available at http://dm.unife.it/gpdt.File in questo prodotto:
Non ci sono file associati a questo prodotto.
I documenti in SFERA sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
