Robust Interactive Learning

TL;DR

This paper explores the theoretical capabilities of class conditional queries in active learning, providing bounds on query complexity under noise and demonstrating adaptive methods for unknown noise rates.

Contribution

It introduces a theoretical framework for class conditional queries in active learning, with bounds and adaptive strategies under noise models.

Findings

Nearly tight bounds on query complexity for agnostic and bounded noise models

Adaptive algorithms that handle unknown noise rates with minimal query loss

Enhanced understanding of class conditional queries' power in noisy environments

Abstract

In this paper we propose and study a generalization of the standard active-learning model where a more general type of query, class conditional query, is allowed. Such queries have been quite useful in applications, but have been lacking theoretical understanding. In this work, we characterize the power of such queries under two well-known noise models. We give nearly tight upper and lower bounds on the number of queries needed to learn both for the general agnostic setting and for the bounded noise model. We further show that our methods can be made adaptive to the (unknown) noise rate, with only negligible loss in query complexity.