Meta-Learning Reliable Priors in the Function Space

TL;DR

This paper introduces F-PACOH, a meta-learning framework that models priors as stochastic processes in function space, improving uncertainty calibration and performance in data-scarce scenarios like Bayesian Optimization.

Contribution

F-PACOH is a novel meta-learning approach that regularizes in function space to produce well-calibrated uncertainty estimates, addressing overconfidence issues in prior methods.

Findings

F-PACOH outperforms existing meta-learners in Bayesian Optimization benchmarks.

The method provides reliable uncertainty estimates in regions with limited training data.

Integration with sequential decision making enhances decision quality.

Abstract

When data are scarce meta-learning can improve a learner's accuracy by harnessing previous experience from related learning tasks. However, existing methods have unreliable uncertainty estimates which are often overconfident. Addressing these shortcomings, we introduce a novel meta-learning framework, called F-PACOH, that treats meta-learned priors as stochastic processes and performs meta-level regularization directly in the function space. This allows us to directly steer the probabilistic predictions of the meta-learner towards high epistemic uncertainty in regions of insufficient meta-training data and, thus, obtain well-calibrated uncertainty estimates. Finally, we showcase how our approach can be integrated with sequential decision making, where reliable uncertainty quantification is imperative. In our benchmark study on meta-learning for Bayesian Optimization (BO), F-PACOH significantly outperforms all other meta-learners and standard baselines.