Time-Aware Latent Space Bayesian Optimization

TL;DR

This paper introduces TALBO, a novel time-aware Bayesian optimization method that adapts to evolving objectives in structured domains like molecular design by incorporating temporal dynamics into the latent space and surrogate model.

Contribution

TALBO is the first approach to integrate time-awareness into latent space Bayesian optimization, effectively handling drifting objectives in molecular design tasks.

Findings

TALBO outperforms existing LSBO methods across various drifting scenarios.

The method remains robust under different drift speeds and design choices.

TALBO is competitive even when objectives do not drift.

Abstract

Latent-space Bayesian optimization (LSBO) extends Bayesian optimization to structured domains, such as molecular design, by searching in the continuous latent space of a generative model. However, most LSBO methods assume a fixed objective, whereas real design campaigns often face temporal drift (e.g., evolving preferences or shifting targets). Bringing time-varying BO into LSBO is nontrivial: drift can affect not only the surrogate, but also the latent search space geometry induced by the representation. We propose Time-Aware Latent-space Bayesian Optimization (TALBO), which incorporates time in both the surrogate and the learned generative representation via a GP-prior variational autoencoder, yielding a latent space aligned as objectives evolve. To evaluate timevarying LSBO systematically, we adapt widely used molecular design tasks to drifting multi-property objectives and introduce metrics tailored to changing targets. Across these benchmarks, TALBO consistently outperforms strong LSBO baselines and remains robust across drift speeds and design choices, while remaining competitive under actually time-invariant objectives.