Time-Aware Latent Space Bayesian Optimization

2026-03-01

AI Summary

This work addresses the performance degradation of latent space Bayesian optimization (LSBO) under time-varying objective functions—such as shifting user preferences or evolving design goals—by proposing Time-Aware Latent Space Bayesian Optimization (TALBO). TALBO is the first method to explicitly model temporal dynamics within the joint learning of the generative and surrogate models, leveraging a Gaussian process prior variational autoencoder (GP-prior VAE) to enable co-evolution of latent representations and the objective function. The authors introduce a benchmark for molecular design with time-varying objectives and demonstrate TALBO’s superiority across diverse multi-objective optimization tasks. The method exhibits robustness under varying drift rates and design choices while maintaining competitive performance in static settings.

Technology Category

• Search and Optimization: Sampling/Simulation-based Search
• Machine Learning: Optimization
• Intelligent Robots: Learning & Optimization for ROB

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs

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.

Problem

• time-varying objectives
• latent-space Bayesian optimization
• temporal drift
• molecular design
• structured domains

Innovation

• Time-Aware Bayesian Optimization
• Latent Space Alignment
• Temporal Drift
• GP-prior VAE
• Molecular Design