MT勉強会/2012

Machine translation at NAIST

• We focus on statistical machine translation, with particular interests in:
  • training & decoding algorithms
  • integration of rich syntactical and semantic information
  • speech translation
• Currently we have two back-to-back meetings: MT Study Group & IWSLT team meeting.
  • The purpose of MT Study Group is to read and discuss research papers.
  • The purpose of IWSLT team meeting is to discuss progress/issues for the IWSLT2012 campaign.

Information

• Date:Thursday
• Time:13:00 - 14:00 IWSLT meeting / 14:00 - 15:00 MT Study Group
• Location: Alternate weekly between Matsumoto-lab seminar room (A707) and Nakamura-lab seminar room.
• Internal page

Schedule 2012

• 4/18 (Wed) 10:00- Kickoff meeting for IWSLT Campaign
  • Learn the KIT system at IWSLT 2011

• 4/26 (Thu) @ Matsumoto-lab: Kickoff meeting for MT Study group

• 4/27 (Fri) 13:30: Invited talk by Taro Watanabe (NICT). Structures in Machine Translation

• 5/10 @ Nakamura-lab: neubig & kevinduh
  • Tutorial: Building a Phrase Based SMT System. Slides in English & Japanese

• 5/17 @ Matsumoto-lab: shuhei-k
  • Dyer et. al. Unsupervised Word Alignment with Arbitrary Features (ACL2011)

• 5/24: takatomo-k
  • Koehn & Hoang. Factored Translation Models (EMNLP2007)

• 5/31 katsuhiko-h
  • Haghighi, DeNero, Klein. Approximate Factoring for A* Search (NAACL2007)

• 6/07 masaya-o
  • Gimpel & Smith. Structured Ramp Loss Minimization for Machine Translation (NAACL2012)

• 6/15 tetsuo-s
  • Simianer et. al. Joint feature selection in distributed stochastic learning for large-scale discriminative training in SMT (ACL2012)

• 6/21 NAACL review
  • kevinduh: Jagarlamudi & Daume, Low-dimensional Discriminative Re-ranking
  • masaya-o: Watanabe, Optimized Online Rank Learning for Machine Translation
  • neubig: Takstrom+, Cross-lingual Word Clusters for Direct Transfer of Linguistic Structure; Ture, Encouraging Consistent Translation Choices
  • shuhei-k: Rush & Petrov, Vine Pruning for Efficient Multi-Pass Dependency Parsing

• 6/28 ACL/EMNLP practice talks
  • kevinduh: Learning to Translate with Multiple Objectives
  • katsuhiko-h: Head-driven Transition-based Parsing with Top-down Prediction
  • neubig: Machine Translation without Words through Substring Alignment
  • neubig: Inducing a Discriminative Parser to Optimize Machine Translation Reordering

• 7/05 tomoki-f
  • Bangalore+, Real-time Incremental Speech-to-Speech Translation of Dialogs (NAACL10)

• 7/19 thichinh-t
  • Genzel, Automatically Learning Source-side Reordering Rules for Large Scale Machine Translation (COLING10)

• 7/26 shuhei-k
  • Burkett+, Joint Parsing and Alignment with Weakly Synchronized Grammars (NAACL10)

• 8/2 neubig & kevinduh (ACL/EMNLP review)
  • Burkett & Klein, Transforming Trees to improve Syntactic Convergence
  • Eidelman, Boyd-Graber, Resnik, Topic Models for Dynamic Translation Model Adaptation

• 8/30 takamoto-k
  • Aguero, Adell, Bonafonte, Prosody generation for speech-to-speech translation

• 9/6 katsuhiko-h
  • Efficient CCG Parsing: A-star vs. Adaptive Supertagging

• 9/13 masaya-o
  •  Tuning as Linear Regression, NAACL 2012

• 9/28 tetsuo-s
  • Exact Decoding of Phrase-Based Translation Models through Lagrangian Relaxation

• 10/4 kevinduh
  • Continuous Space Translation Models with Neural Networks

• 10/11 shuhei-k
  • Exact Decoding of Syntactic Translation Models through Lagrangian Relaxation

• 10/18 neubig
  • A Spectral Algorithm for Learning Hidden Markov Models

• 10/25 masaya-o, miao-y
  • Detection of Non-native Sentences using Machine-translated Training Data

• 11/01 tomoki-f, hiroali-sh
  • Automatic Sentence Segmentation and Punctuation Prediction for Spoken Language Translation

• 11/08 takeyuki-k, takatomo-k
  • A Probabilistic Forest-to-String Model for Language Generation from Typed Lambda Calculus Expressions

• 11/15 isao-ni, kazuya-ko
  • Syntactic Reordering in Preprocessing for Japanese→English Translation: MIT System Description for NTCIR-7 Patent Translation Task

• 11/22 tetsuo-k
  • Optimal Search for Minimum Error Rate Training

• 11/29 kevinduh
  • MERT by Sampling the Translation Lattice

• 12/6 cancel

• 12/13 shuhei-k
  • Adjoining Tree-to-String Translation

• 12/20 neubig
  • Lattice-based Minimum Error Rate Training for Statistical Machine Translation
  • Efficient Minimum Error Rate Training and Minimum Bayes-Risk Decoding for Translation Hypergraphs and Lattices

• 1/10/2013 COLING review
  • kevinduh: Simple and Effective Parameter Tuning for Domain Adaptation of Statistical Machine Translation
  • neubig: Bayesian Language Modelling of German Compounds

• 1/17/2013 COLING review
  • shuhei-k: Stacking of Dependency and Phrase Structure Parsers
  • takatomo-k: Factored Language Model based on Recurrent Neural Network
  • masaya-o: Optimizing for Sentence-Level BLEU+1
  • hiroaki-sh: Active Error Detection and Resolution for Speech-to-Speech Translation (IWSLT2012)
  • isao-ni: Alignment by Bilingual Generation andMonolingual Derivation
  • tomoki-f: Overview of the IWSLT 2011 Evaluation Campaign
  • kazuya-ko: Tree-based Translation without Using Parse Trees

• 1/24/2013 henning-s
  • Continuous Space Language Models using Restricted Boltzmann Machines

• 1/31/2013 takatomo-k, hiroaki-sh
  • Left-to-Right Target Generation for Hierarchical Phrase-based Translation

• 2/7/2013 cancel

• 2/14/2013 isao-ni, kazuya-ko
  • Does more data always yield better translations?

• 2/21/2013 masaya-o, tomoki-f
  • The HIT-LTRC Machine Translation System for IWSLT 2012