Learning from Label Proportions: A Mutual Contamination Framework

TL;DR

This paper introduces a new framework for learning from label proportions using mutual contamination models, providing theoretical foundations and addressing limitations in existing experimental setups.

Contribution

It formulates LLP within MCMs, establishing unbiased losses, generalization bounds, and proposing a new experimental setting for better evaluation.

Findings

Unbiased loss functions for LLP under MCMs

Generalization error bounds for non-iid samples

A new experimental framework for LLP

Abstract

Learning from label proportions (LLP) is a weakly supervised setting for classification in which unlabeled training instances are grouped into bags, and each bag is annotated with the proportion of each class occurring in that bag. Prior work on LLP has yet to establish a consistent learning procedure, nor does there exist a theoretically justified, general purpose training criterion. In this work we address these two issues by posing LLP in terms of mutual contamination models (MCMs), which have recently been applied successfully to study various other weak supervision settings. In the process, we establish several novel technical results for MCMs, including unbiased losses and generalization error bounds under non-iid sampling plans. We also point out the limitations of a common experimental setting for LLP, and propose a new one based on our MCM framework.