Time: 2:15pm-3:30pm Place: Room C415A (basement level), CUNY Graduate
Center (365 5th Ave, between 34str. & 35str)

Speaker: Heng Yu, Chinese Academy of Sciences and City University of New
York

Title: Max-Violation Perceptron and Forced Decoding for Scalable MT
Training

Joint work with Liang Huang (CUNY), Haitao Mi (IBM), and Kai Zhao
(CUNY).

Paper: http://acl.cs.qc.edu/~lhuang/papers/maxforce.pdf

Abstract:

While large-scale discriminative training has triumphed in many NLP
problems, its definite success on machine translation has been largely
elusive. Most recent efforts along this line are not scalable (training
on the small dev set with features from top ∼100 most fre- quent words)
and overly complicated. We instead present a very simple yet
theoretically motivated approach by extending the recent framework of
“violation-fixing perceptron”, using forced decoding to compute the
target derivations. Extensive phrase-based translation experiments on
both Chinese-to-English and Spanish-to-English tasks show substantial
gains in BLEU by up to +2.3/+2.0 on dev/test over MERT, thanks to 20M+
sparse features. This is the first successful effort of large-scale
online discriminative training for MT.

Bio:

Heng Yu is a fifth-year Ph.D. student under Prof. Qun Liu at the
Institute of Computing Technologies, Chinese Academy of Sciences
(CAS/ICT). Since May 2013 he has been visiting the Algorithms for
Computational Linguistics (ACL) Group at CUNY led by Prof. Liang Huang.
His research is on machine translation, in particular, large-scale
discriminative training and syntactic language models.