Poisson Learning: Graph Based Semi-Supervised Learning At Very Low Label Rates

Abstract

We propose a new framework, called Poisson learning, for graph based semi-supervised learning at very low label rates. Poisson learning is motivated by the need to address the degeneracy of Laplacian semi-supervised learning in this regime. The method replaces the assignment of label values at training points with the placement of sources and sinks, and solves the resulting Poisson equation on the graph. The outcomes are provably more stable and informative than those of Laplacian learning. Poisson learning is efficient and simple to implement, and we present numerical experiments showing the method is superior to other recent approaches to semi-supervised learning at low label rates on MNIST, FashionMNIST, and Cifar-10. We also propose a graph-cut enhancement of Poisson learning, called Poisson MBO, that gives higher accuracy and can incorporate prior knowledge of relative class sizes.

Proceedings of the 37th International Conference on Machine Learning, Online, PMLR 119, 2020

Cited by in corpus (11)

• Improved spectral convergence rates for graph Laplacians on epsilon-graphs and k-NN graphs
• Continuum Limit of Lipschitz Learning on Graphs
• Uniform Convergence Rates for Lipschitz Learning on Graphs
• Energy Models for Better Pseudo-Labels: Improving Semi-Supervised Classification with the 1-Laplacian Graph Energy
• Contrastive and Generative Graph Convolutional Networks for Graph-based Semi-Supervised Learning
• PPGN: Physics-Preserved Graph Networks for Real-Time Fault Location in Distribution Systems with Limited Observation and Labels
• Novel Batch Active Learning Approach and Its Application to Synthetic Aperture Radar Datasets
• Model-Change Active Learning in Graph-Based Semi-Supervised Learning
• Ratio convergence rates for Euclidean first-passage percolation: Applications to the graph infinity Laplacian
• Almost exact recovery in noisy semi-supervised learning
• Dual GNNs: Graph Neural Network Learning with Limited Supervision