AI Paper Atlas

Research map / Label-efficient and robust learning / Semi-supervised learning

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

Jeff Calder, Brendan Cook, Matthew Thorpe, Dejan Slepcev

ICML 2020 · Label-efficient and robust learning › Semi-supervised learning · approach: Holistic semi-supervised pipelines · 28 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Add Poisson MBO graph-cut refinement to Poisson learningPoisson learning · MNIST, 1 label/classaccuracy: 90.2 (4.0) -> 96.5 (2.6)Improved
Add Poisson MBO graph-cut refinement to Poisson learningPoisson learning · FashionMNIST, 1 label/classaccuracy: 60.8 (4.6) -> 62.0 (5.7)Improved
Add Poisson MBO graph-cut refinement to Poisson learningPoisson learning · Cifar-10, 1 label/classaccuracy: 40.7 (5.5) -> 41.8 (6.5)Improved
Subtract the estimated tail constant before making label decisionsLaplace/LP · MNIST, 1 label/class (10 labeled images total)accuracy: 16% -> 85.9%Improved
Use Poisson learning instead of Laplace/LPGraph-based semi-supervised classification · MNIST, 1 label/classaccuracy: 16.1 (6.2) -> 90.2 (4.0)Improved
Use Poisson learning instead of Laplace/LPGraph-based semi-supervised classification · FashionMNIST, 1 label/classaccuracy: 18.4 (7.3) -> 60.8 (4.6)Improved
Use Poisson learning instead of Laplace/LPGraph-based semi-supervised classification · Cifar-10, 1 label/classaccuracy: 10.4 (1.3) -> 40.7 (5.5)Improved
Vary the number of neighbors used to construct the graphPoisson learning · Datasets tested, 5 labels/classaccuracyNo clear effect
Compensate for unbalanced training labels in the label decisionPoisson learning · Datasets tested, unbalanced training labelsaccuracyImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy