Model Exploration with Cost-Aware Learning

TL;DR

This paper introduces a cost-aware active learning extension that considers variable costs, enabling learners to explore high-cost regions and outperform traditional methods on standard datasets.

Contribution

It presents a novel extension to active learning that explicitly incorporates non-constant costs, including unknown costs, and introduces the psilon-frugal concept.

Findings

psilon-frugal learners outperform known-cost and random sampling methods.

The approach effectively explores high-cost regions in the sample space.

Demonstrated improvements on a well-known machine learning dataset.

Abstract

We present an extension to active learning routines in which non-constant costs are explicitly considered. This work considers both known and unknown costs and introduces the term \epsilon-frugal for learners that do not only consider minimizing total costs but are also able to explore high cost regions of the sample space. We demonstrate our extension on a well-known machine learning dataset and find that out \epsilon-frugal learners outperform both learners with known costs and random sampling.