On the use of high-order feature propagation in Graph convolution Networks with manifold regularization

Our proposed scheme would lead to more informative GCNs. Using the revisited model, we will conduct several semi-supervised classification experiments on public image datasets containing objects