| Manaal Faruqui |
| Fri 17 Jul 2015, 11:00 - 12:30 |
| Informatics Forum (IF-4.31/4.33) |
If you have a question about this talk, please contact: Diana Dalla Costa (ddallac)
Abstract:
Data-driven learning of distributional word vector representations is a technique of central importance in natural language processing. In this talk, we will explore several questions and their solutions that are aimed at improving and better understanding distributional word vectors. Can word vectors benefit from information stored in semantic lexicons?
Can these word vectors look similar to features typically used in NLP? Do the vector dimensions have certain meaning associated with them or are they uninterpretable? Is it necessary to develop word vectors using distributional context?
Biography:
Manaal Faruqui is a third year PhD student in the Language Technologies Institute at Carnegie Mellon University. During his ongoing PhD, he is working on problems in the areas of representation learning, distributional & lexical semantics and multilingual learning. Prior to joining CMU, he was at the Indian Institute of Technology Kharagpur where he finished his undergraduate in computer science & engineering. He has visited Simon Fraser University (2009), University of Stuttgart (2010), Yahoo labs (2011) and Google Research (2014, 2015) as an internship student. He recently won the best student paper award at NAACL 2015 for his work on incorporating semantic knowledge in word vector representations.