Graph Signal Processing and Deep Learning: Convolution, Pooling, and Topology

TL;DR

This paper investigates how graph signal processing can extend convolutional neural networks to graph-structured data, enhancing performance by leveraging data topology.

Contribution

It introduces methods to adapt CNN components for graph data using GSP and designs graph CNN architectures based on data topology.

Findings

GSP-based CNN components improve performance on graph data

Topology-aware graph CNN architectures outperform traditional methods

Framework bridges deep learning and graph signal processing

Abstract

Deep learning, particularly convolutional neural networks (CNNs), have yielded rapid, significant improvements in computer vision and related domains. But conventional deep learning architectures perform poorly when data have an underlying graph structure, as in social, biological, and many other domains. This paper explores 1)how graph signal processing (GSP) can be used to extend CNN components to graphs in order to improve model performance; and 2)how to design the graph CNN architecture based on the topology or structure of the data graph.