University of Cambridge > > NLIP Seminar Series > Functional Distributional Semantics

Functional Distributional Semantics

Add to your list(s) Download to your calendar using vCal

If you have a question about this talk, please contact Anita Verő.

Vector space models have become popular in computational linguistics, despite the challenges they face when it comes to compositionality, inference, context-dependence, and other issues of interest to semanticists. I will present a probabilistic framework which draws on both formal semantics and recent advances in machine learning. In particular, we can separate predicates from the entities they refer to, and represent predicates not by vectors, but by functions. From a formal semantic point of view, these can be seen as truth-conditional functions, and from a machine learning point of view, these can be seen as classifiers. By defining a probabilistic graphical model incorporating such functions, we can recast many semantic phenomena in terms of Bayesian inference. After describing the framework, and its implementation with neural networks, I will present recent results showing that such model can learn information not captured by vector space models.

This talk is part of the NLIP Seminar Series series.

Tell a friend about this talk:

This talk is included in these lists:

Note that ex-directory lists are not shown.


© 2006-2024, University of Cambridge. Contact Us | Help and Documentation | Privacy and Publicity