# Dependency Parsing with Dilated Iterated Graph CNNs

## TL;DR

This paper introduces DIG-CNNs, a novel graph convolutional neural network architecture that enables efficient, parallelizable dependency parsing on GPUs, matching top neural parsers' performance.

## Contribution

The paper presents a new dilated iterated graph CNN architecture that improves GPU-based dependency parsing efficiency while maintaining competitive accuracy.

## Key findings

- DIG-CNNs perform on par with leading neural parsers.
- The architecture enables efficient end-to-end GPU parsing.
- Experimental results on Penn TreeBank validate the approach.

## Abstract

Dependency parses are an effective way to inject linguistic knowledge into many downstream tasks, and many practitioners wish to efficiently parse sentences at scale. Recent advances in GPU hardware have enabled neural networks to achieve significant gains over the previous best models, these models still fail to leverage GPUs' capability for massive parallelism due to their requirement of sequential processing of the sentence. In response, we propose Dilated Iterated Graph Convolutional Neural Networks (DIG-CNNs) for graph-based dependency parsing, a graph convolutional architecture that allows for efficient end-to-end GPU parsing. In experiments on the English Penn TreeBank benchmark, we show that DIG-CNNs perform on par with some of the best neural network parsers.