Ballpark Learning: Estimating Labels from Rough Group Comparisons

TL;DR

This paper introduces a method to estimate individual labels from coarse, aggregate data with only bounds on label proportions, enabling high-accuracy predictions without labeled examples across various real-world applications.

Contribution

It relaxes previous assumptions by using bounds instead of exact proportions, providing an intuitive algorithm for label estimation from aggregate constraints.

Findings

High accuracy in income prediction using stereotypes

Effective sentiment analysis with minimal information

Successful geographical dialect analysis from aggregate data

Abstract

We are interested in estimating individual labels given only coarse, aggregated signal over the data points. In our setting, we receive sets ("bags") of unlabeled instances with constraints on label proportions. We relax the unrealistic assumption of known label proportions, made in previous work; instead, we assume only to have upper and lower bounds, and constraints on bag differences. We motivate the problem, propose an intuitive formulation and algorithm, and apply our methods to real-world scenarios. Across several domains, we show how using only proportion constraints and no labeled examples, we can achieve surprisingly high accuracy. In particular, we demonstrate how to predict income level using rough stereotypes and how to perform sentiment analysis using very little information. We also apply our method to guide exploratory analysis, recovering geographical differences in twitter dialect.