Abstract: Highlights•We present a distributional metric that measures the accuracy of a subgraph.•We measure distributional differences using only sampling probability models.•We utilize the for subgraph sampling in the course of training of large graphs.
Loading