V Note to prospective students
V My group is conducting research at an international level.
* I have high standards.
* You and your work need to be competitive with graduate students in top universities in the US and Europe (think Stanford, MIT, CMU).
* This is hard work.
* Except for very rare circumstances, you cannot maintain an "outside" hi-tech job while conducting good research.
* You will need to read many research papers.
* You will need to write many research papers.
* There is no guarantee you will finish a thesis "on time".
V You are expected to be independent and self motivated.
* I will push you to do good work, and provide the needed conditions for it.
* But I won't pressure you with deadlines etc.
* If you choose to slack for a week or two or a year, it is up to you.
V I work in an applied field.
V You need to know how to program and like to program. You will program. Quite a lot.
* "I know Java" won't cut it. You will need to learn more.
* You will need to be proficient in a Linux environment.
* You will need to be able to "get things to work".
V My research interests are in applied machine learning (ML), with the main application area being natural language processing (NLP).
* If you want to work with me, you need to take at least 89-680, the Introduction to Natural Language Processing course, or otherwise convince me why you don't need to.
* You need to be curious about at least two of:
- human language;
- getting insights from data;
- statistical learning.
Preferably all three.
V More specific recent interests include:
* Deep learning methods, in particular recurrent neural networks -- why they work and how to use them well.
* Predicting complex linguistic structure.
* Syntax.
* Learning with limited amounts of supervision and large amount of un-annotated data.
* Methods that work well across domains.
* Computational Creativity.
* I will also be happy to discuss your own research interests, after you take the Intro to NLP course.
* I am open to working on interesting ML problems outside of NLP. But you will have to convince me they are interesting.