Joint Feature Selection in Distributed Stochastic Learning for Large-Scale Discriminative Training in SMT

Patrick Simianer, Stefan Riezler, Chris Dyer

Abstract: With a few exceptions, discriminative training in statistical machine translation (SMT) has been content with tuning weights for large feature sets on small development data. Evidence from machine learning indicates that increasing the training sample size results in better prediction. The goal of this paper is to show that this common wisdom can also be brought to bear upon SMT. We deploy local features for SCFG-based SMT that can be read off from rules at runtime, and present a learning algorithm that applies l1/l2 regularization for joint feature selection over distributed stochastic learning processes. We present experiments on learning on 1.5 million training sentences, and show significant improvements over tuning discriminative models on small development sets.