Bayesian method speeds up design searches in huge data spaces
SABLE: Scalable Affine Bayesian Latent-space Exploration
Machine LearningArtificial Intelligence
Summary
Optimizing complex designs like engineering parts or proteins is tough because their possible variations are huge and testing each one is costly. The authors developed SABLE, a technique that uses Bayesian statistics to smartly explore these massive design spaces with fewer tries. It breaks down the search into simpler parts to focus on promising areas quickly, saving computational time dramatically. Their method was tested on simulated problems and found to be both efficient and fast.
What this means in practice
- •For engineering design teams: Speed up trial-and-error searches for complex engineering designs by efficiently exploring huge parameter spaces with fewer tests.
- •For protein engineers: Accelerate optimization of protein designs by managing costly experiments through a scalable search of vast design variations.
Authors
Erik Bodin, Lisa Maria Kreusser, Donney Fan, Henry Moss, Geoff Pleiss, Nicolas Durrande
Abstract
Complex objects such as engineering designs and proteins are often difficult to optimise because their design spaces exhibit high-dimensional structure and objective evaluations are expensive. Generative models provide continuous latent representations, but their latent spaces can have thousands to tens of thousands of dimensions, creating a regime where the latent dimension $D$ greatly exceeds the evaluation budget. We introduce SABLE, a Bayesian optimisation method for high-dimensional latent spaces with an isotropic distribution and limited budget. SABLE uses a Bayesian affine surrogate for prediction and acquisition, treating higher-order structure as surrogate discrepancy. After $N$ observations, the slope update has dimension at most $N-1$, yielding a decomposition of each query into an in-span component optimised by expected improvement (EI) in at most $N$ dimensions and an off-span component governed by a scalar norm. We show that the off-span norm implicitly selected by affine EI can lead to underexploration, and SABLE instead prescribes it using a finite-horizon schedule, without restricting search to a learned subspace. A recompression scheme reduces information to $\mathcal O(M)$ pseudo-observations, bounding refitting cost by $\mathcal O(M^2D+M^3)$ while keeping parameter inference, prediction, and acquisition decisions to tens of milliseconds. Across synthetic benchmarks and latent-space optimisation tasks, SABLE achieves high sample efficiency and reduces decision time by several orders of magnitude at large budgets.