Amittai Axelrod

I've moved!

As of September 2014, I am a postdoc at the University of Maryland, in the Institvte for Advanced Computer Studies (UMIACS).

The core problem I'm interested in working on is that of using existing resources effectively. My primary field is statistical machine translation: building statistical systems to automatically translate text from one human language to another.

I am particularly interested in systems that are task-oriented: travel scenarios, real-time telephone conversations, or particular kinds of documents. These tasks have an implicit distribution of language or linguistic features that differ from the language as a whole. My work quantifies how these distributions are different, and then leverages that difference to build better task-oriented translation systems. The idea is generalizable: Given a natural language processing task, one could use these methods to identify the most relevant data available, and then train a task-specific system.

I defended my dissertation ("Data Selection Methods for Statistical Machine Translation") at the University of Washington in June 2014. I was advised by Xiaodong He at Microsoft Research, and by Mari Ostendorf in the Electrical Engineering department at the University of Washington.

I previously completed an M.Sc. with the Statistical Machine Translation Group led by Philipp Koehn, in the School of Informatics at the University of Edinburgh.

My liege advisor at Edinburgh (magnificent city; smells like toast) was the inestimable Miles Osborne. A copy of my M.Sc. thesis can be found here.


Applications of Data Selection via Cross-Entropy Difference for Real-World Statistical Machine Translation.
Amittai Axelrod, Qingjun Li, Will Lewis.
IWSLT (International Workshop on Spoken Language Translation), 2012.

An automated approach for annual layer counting in ice cores.
Mai Winstrup, A. Svensson, S. O. Rasmussen, O. Winther, E. J. Steig, Amittai Axelrod.
Climate of the Past Discussions 8 (4) p. 2519-2555, 2012. (link)

New methods and evaluation experiments on translating TED talks in the IWSLT benchmark
Amittai Axelrod, Xiaodong He, Li Deng, Alex Acero, Mei-Yuh Hwang
ICASSP (International Conference on Acoustics, Speech, and Signal Processing), 2012. (pdf)

Domain Adaptation Via Pseudo In-Domain Data Selection
Amittai Axelrod, Xiaodong He, Jianfeng Gao
EMNLP (Empirical Methods in Natural Language Processing), 2011. (pdf)

Application of statistical machine translation to public health information: a feasibility study.
Katrin Kirchhoff, Anne M. Turner, Amittai Axelrod, Francisco Saavedra.
JAMIA (Journal of the American Medical Informatics Association) 18 (4) p. 473-8, 2011. (link)

The MSR System for IWSLT 2011 Evaluation
Xiaodong He, Amittai Axelrod, Li Deng, Alex Acero, Mei-yuh Hwang, Alisa Nguyen, Andrew Wang, Xiahui Huang.
IWSLT (International Workshop on Spoken Language Translation), 2011. (

The MSRA Machine Translation System for IWSLT 2010
Chi-Ho Li, Nan Duan, Yinggong Zhao, Shujie Liu, Lei Cui, Mei-Yuh Hwang, Amittai Axelrod, Jianfeng Gao, Yaodong Zhang, Li Deng
IWSLT (International Workshop on Spoken Language Translation), 2010. (pdf)

The University of Washington Machine Translation System for ACL WMT 2008
Amittai Axelrod, Mei Yang, Kevin Duh, Katrin Kirchhoff.
WMT (Workshop on Statistical Machine Translation), 2008.

Edinburgh System Description for the 2006 TC-STAR Spoken Language Translation Evaluation.
Abhishek Arun, Amittai Axelrod, Philipp Koehn.
TC-Star Evaluation Workshop, 2006.

Edinburgh System Description for the 2005 IWSLT Speech Translation Evaluation.
Philipp Koehn, Amittai Axelrod, Alexandra Birch-Mayne, Chris Callison-Burch, Miles Osborne, David Talbot.
IWSLT (International Workshop on Spoken Language Translation), 2005. (pdf)

Edinburgh System Description for the 2005 NIST MT Evaluation.
Philipp Koehn, Amittai Axelrod, Alexandra Birch Mayne, Chris Callison-Burch, Miles Osborne, David Talbot, and Michael White.
MT Eval Workshop, 2005.

Older stuff:

During my second internship at Microsoft Research, I wrote what appears to have been the first step-by-step guide to compiling and installing the Moses MT framework on a Windows machine:
Moses on Win7 (last updated 15 July 2011)
The guide is now out-of-date, as Moses has been updated to install smoothly out-of-the-box on Windows, but it's an example of how I document things.

Last updated: Sat Jun 28 13:03:51 PDT 2014