Approximation-Aware Dependency Parsing by Belief Propagation

TL;DR

This paper improves a dependency parser by training it to account for approximation errors in belief propagation, leading to higher accuracy with fewer iterations.

Contribution

It introduces a method to train a dependency parser that considers higher-order interactions and compensates for belief propagation approximation errors.

Findings

Higher accuracy achieved with fewer belief propagation iterations.

Training method effectively compensates for approximation errors.

Parser considers higher-order interactions while maintaining O(n^3) complexity.

Abstract

We show how to train the fast dependency parser of Smith and Eisner (2008) for improved accuracy. This parser can consider higher-order interactions among edges while retaining O(n^3) runtime. It outputs the parse with maximum expected recall -- but for speed, this expectation is taken under a posterior distribution that is constructed only approximately, using loopy belief propagation through structured factors. We show how to adjust the model parameters to compensate for the errors introduced by this approximation, by following the gradient of the actual loss on training data. We find this gradient by back-propagation. That is, we treat the entire parser (approximations and all) as a differentiable circuit, as Stoyanov et al. (2011) and Domke (2010) did for loopy CRFs. The resulting trained parser obtains higher accuracy with fewer iterations of belief propagation than one trained by conditional log-likelihood.