Word alignments are useful for a number of downstream tasks, but NMT has largely forgotten about them. We reintroduce old notions of supervised alignment and present a neural model for aligning words between a sentence and its translation. We apply our aligner to dataset projection for NER.
I am a second-year Ph.D. student at the Center for Language and Speech Processing supervised by Benjamin Van Durme, where I am working on broad questions in computational semantics. I am especially interested in working in cross-lingual settings and with low-resource languages. Before starting my Ph.D., I received my B.A.&Sc. with First Class Honours in Cognitive Science from McGill University, focusing in computer science and linguistics. While at McGill, I worked as a research assistant at the Montreal Language Modeling Lab (MLML), now MCQLL. I wrote my honours thesis (supervised by Timothy O'Donnell) on a variational inference algorithm for a model of language acquisition.
I describe a variational inference scheme and novel phonological system for the Unsupervised Lexicon Discovery model presented by Lee et al. (2015)
A poster describing the work laid out in my thesis, presented at the Montreal AI Symposium (2017) with Emily Kellison-Linn.
A poster I was invited to present at the McGill Arts Undergraduate Research Internship Research Event (2017) on my experience as an ARIA (Arts Research Internship Award) recipient.