Diffusion-Convolutional Neural Networks
This addresses graph-based learning tasks like node classification, offering a novel approach with polynomial-time efficiency and GPU implementation, though it is incremental as it builds on existing neural network and graph methods.
The paper tackled the problem of node classification on graph-structured data by introducing diffusion-convolutional neural networks (DCNNs), which use a diffusion-convolution operation to learn invariant representations, and demonstrated that DCNNs outperform probabilistic relational models and kernel-on-graph methods in experiments on real datasets.
We present diffusion-convolutional neural networks (DCNNs), a new model for graph-structured data. Through the introduction of a diffusion-convolution operation, we show how diffusion-based representations can be learned from graph-structured data and used as an effective basis for node classification. DCNNs have several attractive qualities, including a latent representation for graphical data that is invariant under isomorphism, as well as polynomial-time prediction and learning that can be represented as tensor operations and efficiently implemented on the GPU. Through several experiments with real structured datasets, we demonstrate that DCNNs are able to outperform probabilistic relational models and kernel-on-graph methods at relational node classification tasks.