Sciweavers

Share
ICML
2010
IEEE

Non-Local Contrastive Objectives

9 years 9 months ago
Non-Local Contrastive Objectives
Pseudo-likelihood and contrastive divergence are two well-known examples of contrastive methods. These algorithms trade off the probability of the correct label with the probabilities of other "nearby" instantiations. In this paper we explore more general types of contrastive objectives, which trade off the probability of the correct label against an arbitrary set of other instantiations. We prove that a large class of contrastive objectives are consistent with maximum likelihood, even for finite amounts of data. This result generalizes asymptotic consistency for pseudo-likelihood. The proof gives significant insight into contrastive objectives, suggesting that they enforce (soft) probabilityratio constraints between pairs of instantiations. Based on this insight, we propose Contrastive Constraint Generation (CCG), an iterative constraint-generation style algorithm that allows us to learn a log-linear model using only MAP inference. We evaluate CCG on a scene classification ...
David Vickrey, Cliff Chiung-Yu Lin, Daphne Koller
Added 09 Nov 2010
Updated 09 Nov 2010
Type Conference
Year 2010
Where ICML
Authors David Vickrey, Cliff Chiung-Yu Lin, Daphne Koller
Comments (0)
books