Semi-supervised Learning on Large Graphs: is Poisson Learning a Game-Changer?

TL;DR

This paper investigates Poisson learning for semi-supervised learning on large graphs, analyzing its effectiveness compared to Laplace-based methods and finding it does not resolve global information loss issues.

Contribution

The study clarifies that Poisson learning is essentially Laplace regularization with thresholding and does not address the global information loss problem.

Findings

Poisson learning is equivalent to Laplace regularization with thresholding.

It cannot overcome the global information loss problem in large graphs.

Poisson learning is not a game-changer for semi-supervised learning on large graphs.

Abstract

We explain Poisson learning on graph-based semi-supervised learning to see if it could avoid the problem of global information loss problem as Laplace-based learning methods on large graphs. From our analysis, Poisson learning is simply Laplace regularization with thresholding, cannot overcome the problem.