Survey of Graph Neural Networks and Applications

The advance of deep learning has shown great potential in applications (speech, image, and video classification). In these applications, models are trained by datasets Euclidean space with fixed dimensions sequences. Nonetheless, the rapidly increasing demands on analyzing non-Euclidean require additional research. Generally speaking, finding relationships elements representing such as weighted graphs consisting vertices edges is a viable way space. However, graph-based dataset challenging problem existing models. To address this issue, graph neural networks (GNNs) leverage spectral spatial strategies to extend implement convolution operations Based theory, number enhanced GNNs proposed deal datasets. study, we first review artificial GNNs. We then present ways introduce GNN-based approaches based strategies. Furthermore, discuss some typical Internet Things (IoT) that employ strategies, followed limitations current stage.