Newest papers

SABLE: Scalable Affine Bayesian Latent-space Exploration

Erik Bodin, Lisa Maria Kreußer, Donney Fan, Henry B. Moss, Geoff Pleiss, Nicolas Durrande · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.31495 · cited by 0 works (OpenAlex) · open-access PDF

Portuguese

Os modelos generativos fornecem representações paramétricas latentes contínuas, mas seus espaços latentes podem ter milhares a dezenas de milhares de dimensões de ordem, criando um regime onde a dimensão latente $D$ ultrapassa consideravelmente o orçamento de avaliação. Introduzimos SABLE, um método de otimização da inclinação Bayesiana para espaços latentes de alta dimensão com uma distribuição isotópica contínua e orçamento limitado. SABLE usa um substituto de afinidade Bayesiana para previsão e aquisição, tratando a estrutura de ordem superior como uma discrepância de substituição. Após as observações de $N$, a atualização da inclinação tem uma dimensão maior que $N-1$, resultando em uma decomposição de cada esquema latente em um compon

Machine translation (Meta M2M100); the English original follows.

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.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Bounding overwatch, Bayesian probability, Affine transformation, Dimension (graph theory), Computer science, Mathematics

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange