Loss-Driven Bayesian Active Learning

TL;DR

This paper introduces a loss-driven Bayesian active learning method that directly optimizes data acquisition for specific decision losses, improving predictive performance across tasks.

Contribution

It presents a novel framework allowing flexible, loss-specific data acquisition in Bayesian active learning, with analytic solutions for weighted Bregman divergence losses.

Findings

Reduces test losses compared to existing methods in regression and classification.

Provides a unified approach for different loss functions in active learning.

Abstract

The central goal of active learning is to gather data that maximises downstream predictive performance, but popular approaches have limited flexibility in customising this data acquisition to different downstream problems and losses. We propose a rigorous loss-driven approach to Bayesian active learning that allows data acquisition to directly target the loss associated with a given decision problem. In particular, we show how any loss can be used to derive a unique objective for optimal data acquisition. Critically, we then show that any loss taking the form of a weighted Bregman divergence permits analytic computation of a central component of its corresponding objective, making the approach applicable in practice. In regression and classification experiments with a range of different losses, we find our approach reduces test losses relative to existing techniques.