TreeRNN: Topology-Preserving Deep GraphEmbedding and Learning

TL;DR

TreeRNN introduces a novel approach to graph learning by converting graphs into directed trees via BFS, transforming them into image representations, and applying a 2D RNN for effective classification, achieving competitive results.

Contribution

The paper presents a new graph-to-image projection scheme and a 2D RNN architecture for topology-preserving graph classification, addressing limitations of existing local pattern methods.

Findings

Achieved comparable accuracy with state-of-the-art on multiple datasets.

Demonstrated the effectiveness of graph-to-image projection for CNN and RNN models.

Validated the proposed method on several graph classification benchmarks.

Abstract

General graphs are difficult for learning due to their irregular structures. Existing works employ message passing along graph edges to extract local patterns using customized graph kernels, but few of them are effective for the integration of such local patterns into global features. In contrast, in this paper we study the methods to transfer the graphs into trees so that explicit orders are learned to direct the feature integration from local to global. To this end, we apply the breadth first search (BFS) to construct trees from the graphs, which adds direction to the graph edges from the center node to the peripheral nodes. In addition, we proposed a novel projection scheme that transfer the trees to image representations, which is suitable for conventional convolution neural networks (CNNs) and recurrent neural networks (RNNs). To best learn the patterns from the graph-tree-images, we propose TreeRNN, a 2D RNN architecture that recurrently integrates the image pixels by rows and columns to help classify the graph categories. We evaluate the proposed method on several graph classification datasets, and manage to demonstrate comparable accuracy with the state-of-the-art on MUTAG, PTC-MR and NCI1 datasets.