Correlating neural and symbolic representations of language

Grzegorz Chrupała, Afra Alishahi

Abstract:

Analysis methods which enable us to better understand the
representations and functioning of neural models of language are
increasingly needed as deep learning becomes the dominant approach
in NLP. Here we present two methods based on Representational
Similarity Analysis (RSA) and Tree Kernels (TK) which allow us to
directly quantify how strongly the information encoded in neural
activation patterns corresponds to information represented by
symbolic structures such as syntax trees. We first validate our
methods on the case of a simple synthetic language for arithmetic
expressions with clearly defined syntax and semantics, and show that
they exhibit the expected pattern of results. We then apply our methods to
correlate neural representations of English sentences with their
constituency parse trees.

Keywords: natural language processing, language modeling, analysis methods for neural networks, representational similarity analysis, tree kernels

TL;DR: Two methods based on Representational Similarity Analysis (RSA) and Tree Kernels (TK) which directly quantify how strongly information encoded in neural activation patterns corresponds to information represented by symbolic structures.