Scaling Results in Lp Regularised Semi-Supervised LearningMS17

The semi-supervised learning problem is to assign labels to a dataset given a small number of training labels. We consider random geometric graphs and a Lp finite difference regulariser. In the talk I will discuss the asymptotic behaviour when the number of unlabelled data points increases whilst the number of training labels remains fixed. We show a delicate interplay between the regularizing nature of the functionals considered and the nonlocality inherent to the graph constructions.

This presentation is part of Minisymposium “MS17 - Discrete-to-continuum graphical methods for large-data clustering, classification and segmentation
organized by: Matthew Thorpe (University of Cambridge) , Luca Calatroni (CMAP, École Polytechnique CNRS) , Daniel Tenbrinck (University of Münster) .

Matthew Thorpe (University of Cambridge)
machine learning, partial differential equation models