Sampling for network function learning
This addresses the challenge of scalability in network analysis for researchers and practitioners dealing with large-scale graph data.
The paper tackles the problem of learning network functions from large or dynamic graphs by proposing a graph sampling approach, which allows for efficient processing when edges are unknown or the graph is too large to handle entirely.
Given a valued graph, where both the nodes and the edges of the graph are associated with one or several values, any network function for a given node must be defined in terms of that node and its connected nodes in the graph. Generally, applying the same definition to the whole graph or any given subgraph of it would result in systematically different network functions. In this paper we consider the feasibility of graph sampling approach to network function learning, as well as the corresponding learning methods based on the sample graphs. This can be useful either when the edges are unknown to start with or the graph is too large (or dynamic) to be processed entirely.