A directed graph convolutional neural network for edge-structured signals in link-fault detection

The growing interest in graph deep learning has led to a surge of research focusing on various characteristics graph-structured data. Directed graphs have generally been treated as incidental definitions the more general class undirected graphs. implicit imbalance some problems also proves difficult tackle. Moreover, body work begun grow that considers how learn signals structured edges In this paper, we propose directed convolutional neural network (DGCNN), and describe simple way mitigate inherent model is applied edge-structured from datacenter simulations using structure linegraph represent second-order its underlying graph. We demonstrate DGCNN’s improves over models other by applying our locating link-faults simulation.