Directed Acyclic Graph Neural Networks

TL;DR

This paper introduces DAGNN, a novel graph neural network architecture tailored for directed acyclic graphs that incorporates partial ordering, outperforming existing models on various DAG datasets.

Contribution

DAGNN is a new GNN framework that explicitly models partial orderings in DAGs, unifying and extending previous architectures with improved performance.

Findings

DAGNN outperforms existing DAG and general GNN architectures.

Ablation studies highlight the importance of key components in DAGNN.

DAGNN effectively processes diverse DAG datasets like source code and probabilistic models.

Abstract

Graph-structured data ubiquitously appears in science and engineering. Graph neural networks (GNNs) are designed to exploit the relational inductive bias exhibited in graphs; they have been shown to outperform other forms of neural networks in scenarios where structure information supplements node features. The most common GNN architecture aggregates information from neighborhoods based on message passing. Its generality has made it broadly applicable. In this paper, we focus on a special, yet widely used, type of graphs -- DAGs -- and inject a stronger inductive bias -- partial ordering -- into the neural network design. We propose the \emph{directed acyclic graph neural network}, DAGNN, an architecture that processes information according to the flow defined by the partial order. DAGNN can be considered a framework that entails earlier works as special cases (e.g., models for trees and models updating node representations recurrently), but we identify several crucial components that prior architectures lack. We perform comprehensive experiments, including ablation studies, on representative DAG datasets (i.e., source code, neural architectures, and probabilistic graphical models) and demonstrate the superiority of DAGNN over simpler DAG architectures as well as general graph architectures.