Graph variable estimation models, devices and methods that convolve with in-graph relationships that depend on edge features

摘要

PROBLEM TO BE SOLVED: To provide a graph variable estimation model capable of more reliably incorporating an in-graph relationship inherent in a graph and estimating information related to the graph. SOLUTION: This graph variable estimation model takes graph information related to a graph as input, convolves the graph information in a plurality of intermediate layers having virtual nodes corresponding to the nodes of the graph, and estimates the objective variable for the graph. It is a model to do. Here, at least one in-graph relationship is set corresponding to each of at least one value range for the value taken by the edge feature amount representing the edge feature in the graph, and each virtual of one intermediate layer. The node convolves the node features representing the characteristics of the virtual node corresponding to the in-graph relationship in the intermediate layer in the previous stage for each in-graph relationship included in the set of in-graph relationships including at least one in-graph relationship. , Generates node features that represent their own features. [Selection diagram] Fig. 1