JUMBO: Scalable Multi-task Bayesian Optimization using Offline Data
This work addresses the problem of efficient black-box function optimization for researchers and practitioners in fields like machine learning and engineering, offering a scalable solution with theoretical guarantees, though it builds incrementally on prior MBO methods.
The paper tackles the scalability limitations of Multi-task Bayesian Optimization (MBO) with large offline datasets by proposing JUMBO, which uses a dual Gaussian Process approach to dynamically balance online and offline data, achieving significant performance improvements in hyper-parameter optimization and automated circuit design.
The goal of Multi-task Bayesian Optimization (MBO) is to minimize the number of queries required to accurately optimize a target black-box function, given access to offline evaluations of other auxiliary functions. When offline datasets are large, the scalability of prior approaches comes at the expense of expressivity and inference quality. We propose JUMBO, an MBO algorithm that sidesteps these limitations by querying additional data based on a combination of acquisition signals derived from training two Gaussian Processes (GP): a cold-GP operating directly in the input domain and a warm-GP that operates in the feature space of a deep neural network pretrained using the offline data. Such a decomposition can dynamically control the reliability of information derived from the online and offline data and the use of pretrained neural networks permits scalability to large offline datasets. Theoretically, we derive regret bounds for JUMBO and show that it achieves no-regret under conditions analogous to GP-UCB (Srinivas et. al. 2010). Empirically, we demonstrate significant performance improvements over existing approaches on two real-world optimization problems: hyper-parameter optimization and automated circuit design.