Unifying local and non-local signal processing with graph CNNs
This work addresses the challenge of integrating diverse signal processing techniques in graph-based machine learning, but appears incremental as it builds upon existing graph CNN methods.
The paper tackles the problem of unifying local and non-local signal processing on graphs by proposing a convolutional neural network framework that adapts to graph structure changes, resulting in a novel method for style transfer.
This paper deals with the unification of local and non-local signal processing on graphs within a single convolutional neural network (CNN) framework. Building upon recent works on graph CNNs, we propose to use convolutional layers that take as inputs two variables, a signal and a graph, allowing the network to adapt to changes in the graph structure. In this article, we explain how this framework allows us to design a novel method to perform style transfer.