Abstract
This review provides a comprehensive overview of the state-of-the-art methods of graph-based networks from a deep learning perspective. Graph networks provide a generalized form to exploit non-euclidean space data. A graph can be visualized as an aggregation of nodes and edges without having any order. Data-driven architecture tends to follow a fixed neural network trying to find the pattern in feature space. These strategies have successfully been applied to many applications for euclidean space data. Since graph data in a non-euclidean space does not follow any kind of order, these solutions can be applied to exploit the node relationships. Graph Neural Networks (GNNs) solve this problem by exploiting the relationships among graph data. Recent developments in computational hardware and optimization allow graph networks possible to learn the complex graph relationships. Graph networks are therefore being actively used to solve many problems including protein interface, classification, and learning representations of fingerprints. To encapsulate the importance of graph models, in this paper, we formulate a systematic categorization of GNN models according to their applications from theory to real-life problems and provide a direction of the future scope for the applications of graph models as well as highlight the limitations of existing graph networks.
Highlights
The graph is a variant of data structure that learns the relationship between nodes and explores the relationship among these nodes
The classification of spectral and spatial based methods depends on the types of the convolution operation. As it only focuses on convolution filters, other types of network structures for graph models are overlooked
DATA-DRIVEN METHODS FOR GRAPH EXPLOITATION In this sub-section, we provide a number of graph-data exploitation methods for non-euclidean space data exploration
Summary
The graph is a variant of data structure that learns the relationship between nodes and explores the relationship among these nodes. The Graph Inception model uses this meta path information propagation system for a better understanding of graph structure data [11] This network handles the heterogeneous graph converting it into sub-graph for the information propagation and adds the results from different sub-graphs for node feature learning process. The classification of spectral and spatial based methods depends on the types of the convolution operation As it only focuses on convolution filters, other types of network structures for graph models are overlooked. This makes the review limited to only GCN models. In contrast to the spatial and spectral-based graph networks, an application-oriented graph-based taxonomy has been proposed focusing on both theoretical and reallife processes.
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call