Statistics Papers

Document Type

Working Paper

Date of this Version

7-2014

Publication Source

Journal of Machine Learning Research

Volume

15

Start Page

2399

Last Page

2449

Abstract

We introduce a spectral learning algorithm for latent-variable PCFGs (Matsuzaki et al., 2005; Petrov et al., 2006). Under a separability (singular value) condition, we prove that the method provides statistically consistent parameter estimates. Our result rests on three theorems: the first gives a tensor form of the inside- outside algorithm for PCFGs; the second shows that the required tensors can be estimated directly from training examples where hidden-variable values are missing; the third gives a PAC-style convergence bound for the estimation method.

Comments

At the time of publication, author Dean P. Foster was affiliated with Yahoo! Labs. Currently, he is a faculty member at the Statistics Department at the University of Pennsylvania.

Keywords

latent-variable PCFGs, spectral learning algorithms

Share

COinS
 

Date Posted: 27 November 2017

This document has been peer reviewed.