Attentional Multilabel Learning over Graphs: A Message Passing Approach

TL;DR

This paper introduces GAML, a novel graph neural network that models label relations and substructure interactions through message passing and attention, significantly improving multilabel classification on graph data.

Contribution

GAML is the first model to integrate label nodes with input graphs using message passing and attention, capturing label-substructure relations effectively.

Findings

GAML outperforms existing methods on various datasets.

The model scales linearly with labels and graph size.

GAML provides interpretable visualizations of label relations.

Abstract

We address a largely open problem of multilabel classification over graphs. Unlike traditional vector input, a graph has rich variable-size substructures which are related to the labels in some ways. We believe that uncovering these relations might hold the key to classification performance and explainability. We introduce GAML (Graph Attentional Multi-Label learning), a novel graph neural network that can handle this problem effectively. GAML regards labels as auxiliary nodes and models them in conjunction with the input graph. By applying message passing and attention mechanisms to both the label nodes and the input nodes iteratively, GAML can capture the relations between the labels and the input subgraphs at various resolution scales. Moreover, our model can take advantage of explicit label dependencies. It also scales linearly with the number of labels and graph size thanks to our proposed hierarchical attention. We evaluate GAML on an extensive set of experiments with both graph-structured inputs and classical unstructured inputs. The results show that GAML significantly outperforms other competing methods. Importantly, GAML enables intuitive visualizations for better understanding of the label-substructure relations and explanation of the model behaviors.