PhD position at University of Nantes: Grammatical inference of probabilistic context-free grammars

This position is offered at the University of Nantes in the LINA research lab, in the natural language processing group and is funded by the region “Pays de Loire”.

Grammatical inference is about learning grammars or automata for a language, given partial information, typically strings, about the language. If the situation for finite state machines is fairly well understood (which doesn’t mean solved!) the situation for context-free languages and grammars is still far from clear. Some algorithms exist, but there is room for improvement, in a problem which is important for many applications, of which natural language processing. Probabilistic context-free grammars are context-free grammars in which each rule has a probability of being used: these grammars allow to define interesting distributions over sets of strings. The goal of this PhD is to reconsider different strategies in order to study the learnability of probabilistic context-free grammars, to propose new algorithms and to experiment these on NLP applications.

Applications are invited for a fully funded studentship on the topic stated above.

Qualifications required:
– Master’s in Computer Science or a related field.
– Good mathematical understanding.
– High motivation for research.
– Capability of working in an autonomous way.
– Good programming skills.
– Good communication skills in English, both in written and oral form.
– Experience and detailed knowledge of machine learning, algorithmics, formal language theory and/or statistics will be an asset.
Applicants should submit a complete CV and recommendation letters to cdlhuniv-nantes fr.

Applications are open until the end of the Spring 2011.

The PhD will start (unless otherwise negociated) in September 2011, and the student will be funded during 3 years.

Salary: Standard French PhD scholarship (20K€ p.a.). This salary is compatible with other schemes allowing the student to be able to teach (for example). For more information about these questions, please contact Colin de la Higuera.