Active Tolerant Testing

TL;DR

This paper introduces the first algorithms for tolerant testing of certain classes in the active model, enabling error estimation with minimal label queries and independent of VC-dimension, with applications to performance estimation and hyperparameter tuning.

Contribution

It provides novel algorithms for tolerant testing in the active model for classes like unions of intervals, extending prior non-tolerant work and enabling efficient error and performance estimation.

Findings

Label query complexity is independent of VC-dimension.

Algorithms accurately estimate error rates with few labels.

Applications include hyperparameter tuning and performance estimation.

Abstract

In this work, we give the first algorithms for tolerant testing of nontrivial classes in the active model: estimating the distance of a target function to a hypothesis class C with respect to some arbitrary distribution D, using only a small number of label queries to a polynomial-sized pool of unlabeled examples drawn from D. Specifically, we show that for the class D of unions of d intervals on the line, we can estimate the error rate of the best hypothesis in the class to an additive error epsilon from only $O(\frac{1}{\epsilon^6}\log \frac{1}{\epsilon})$ label queries to an unlabeled pool of size $O(\frac{d}{\epsilon^2}\log \frac{1}{\epsilon})$. The key point here is the number of labels needed is independent of the VC-dimension of the class. This extends the work of Balcan et al. [2012] who solved the non-tolerant testing problem for this class (distinguishing the zero-error case from the case that the best hypothesis in the class has error greater than epsilon). We also consider the related problem of estimating the performance of a given learning algorithm A in this setting. That is, given a large pool of unlabeled examples drawn from distribution D, can we, from only a few label queries, estimate how well A would perform if the entire dataset were labeled? We focus on k-Nearest Neighbor style algorithms, and also show how our results can be applied to the problem of hyperparameter tuning (selecting the best value of k for the given learning problem).