Specializing Word Embeddings (for Parsing) by Information Bottleneck

TL;DR

This paper introduces a fast variational information bottleneck method to compress pre-trained word embeddings, enhancing parsing accuracy by retaining only task-relevant information, either as discrete tags or continuous vectors.

Contribution

It proposes a novel, efficient nonlinear compression technique for word embeddings that improves parsing performance across multiple languages.

Findings

Discrete tags capture most of the information in POS tags.

Moderate compression improves parser accuracy in 8 of 9 languages.

The method outperforms simple dimensionality reduction.

Abstract

Pre-trained word embeddings like ELMo and BERT contain rich syntactic and semantic information, resulting in state-of-the-art performance on various tasks. We propose a very fast variational information bottleneck (VIB) method to nonlinearly compress these embeddings, keeping only the information that helps a discriminative parser. We compress each word embedding to either a discrete tag or a continuous vector. In the discrete version, our automatically compressed tags form an alternative tag set: we show experimentally that our tags capture most of the information in traditional POS tag annotations, but our tag sequences can be parsed more accurately at the same level of tag granularity. In the continuous version, we show experimentally that moderately compressing the word embeddings by our method yields a more accurate parser in 8 of 9 languages, unlike simple dimensionality reduction.