eduzhai > Applied Sciences > Engineering >

TreeRNN Topology-Preserving Deep GraphEmbedding and Learning

  • Save

... pages left unread,continue reading

Document pages: 7 pages

Abstract: General graphs are difficult for learning due to their irregular structures.Existing works employ message passing along graph edges to extract localpatterns using customized graph kernels, but few of them are effective for theintegration of such local patterns into global features. In contrast, in thispaper we study the methods to transfer the graphs into trees so that explicitorders are learned to direct the feature integration from local to global. Tothis end, we apply the breadth first search (BFS) to construct trees from thegraphs, which adds direction to the graph edges from the center node to theperipheral nodes. In addition, we proposed a novel projection scheme thattransfer the trees to image representations, which is suitable for conventionalconvolution neural networks (CNNs) and recurrent neural networks (RNNs). Tobest learn the patterns from the graph-tree-images, we propose TreeRNN, a 2DRNN architecture that recurrently integrates the image pixels by rows andcolumns to help classify the graph categories. We evaluate the proposed methodon several graph classification datasets, and manage to demonstrate comparableaccuracy with the state-of-the-art on MUTAG, PTC-MR and NCI1 datasets.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×