Cost-Sensitive Multi-Fidelity Bayesian Optimization with Transfer of Learning Curve Extrapolation

TL;DR

This paper introduces a cost-sensitive multi-fidelity Bayesian Optimization framework that uses transfer learning for learning curve extrapolation, enabling efficient hyperparameter tuning with early stopping based on user-defined utility functions.

Contribution

It proposes a novel acquisition function and stopping criterion that optimize utility, and enhances learning curve extrapolation with transfer learning for better sample efficiency.

Findings

Outperforms existing multi-fidelity BO and transfer-BO methods.

Achieves better cost-performance trade-offs in hyperparameter optimization.

Effectively captures configuration correlations for improved surrogate modeling.

Abstract

In this paper, we address the problem of cost-sensitive multi-fidelity Bayesian Optimization (BO) for efficient hyperparameter optimization (HPO). Specifically, we assume a scenario where users want to early-stop the BO when the performance improvement is not satisfactory with respect to the required computational cost. Motivated by this scenario, we introduce utility, which is a function predefined by each user and describes the trade-off between cost and performance of BO. This utility function, combined with our novel acquisition function and stopping criterion, allows us to dynamically choose for each BO step the best configuration that we expect to maximally improve the utility in future, and also automatically stop the BO around the maximum utility. Further, we improve the sample efficiency of existing learning curve (LC) extrapolation methods with transfer learning, while successfully capturing the correlations between different configurations to develop a sensible surrogate function for multi-fidelity BO. We validate our algorithm on various LC datasets and found it outperform all the previous multi-fidelity BO and transfer-BO baselines we consider, achieving significantly better trade-off between cost and performance of BO.