Non-Local Feature Aggregation on Graphs via Latent Fixed Data Structures

TL;DR

This paper introduces a novel global feature aggregation method for Graph Neural Networks using Latent Fixed Data Structures, enabling more effective and flexible aggregation compared to traditional pooling methods.

Contribution

The paper proposes a new approach employing various LFDSs for feature aggregation in GNNs, enhancing flexibility and performance over existing methods.

Findings

Achieves competitive or better results than traditional pooling methods.

Linear computational complexity with respect to input graph size.

Introduces multiple LFDSs including loop, tensor, sequence, and data-driven graphs.

Abstract

In contrast to image/text data whose order can be used to perform non-local feature aggregation in a straightforward way using the pooling layers, graphs lack the tensor representation and mostly the element-wise max/mean function is utilized to aggregate the locally extracted feature vectors. In this paper, we present a novel approach for global feature aggregation in Graph Neural Networks (GNNs) which utilizes a Latent Fixed Data Structure (LFDS) to aggregate the extracted feature vectors. The locally extracted feature vectors are sorted/distributed on the LFDS and a latent neural network (CNN/GNN) is utilized to perform feature aggregation on the LFDS. The proposed approach is used to design several novel global feature aggregation methods based on the choice of the LFDS. We introduce multiple LFDSs including loop, 3D tensor (image), sequence, data driven graphs and an algorithm which sorts/distributes the extracted local feature vectors on the LFDS. While the computational complexity of the proposed methods are linear with the order of input graphs, they achieve competitive or better results.