Guided Trajectory Generation with Diffusion Models for Offline Model-based Optimization

TL;DR

This paper introduces a novel diffusion-based generative approach for offline model-based optimization, producing high-quality trajectories to identify optimal designs in high-dimensional black-box functions without online evaluation.

Contribution

It proposes a new conditional diffusion model that generates trajectories toward high-scoring regions, overcoming limitations of prior methods in offline optimization.

Findings

Outperforms baseline methods on Design-Bench tasks

Generates diverse high-quality solutions beyond dataset scope

Effectively guides exploration of high-scoring regions

Abstract

Optimizing complex and high-dimensional black-box functions is ubiquitous in science and engineering fields. Unfortunately, the online evaluation of these functions is restricted due to time and safety constraints in most cases. In offline model-based optimization (MBO), we aim to find a design that maximizes the target function using only a pre-existing offline dataset. While prior methods consider forward or inverse approaches to address the problem, these approaches are limited by conservatism and the difficulty of learning highly multi-modal mappings. Recently, there has been an emerging paradigm of learning to improve solutions with synthetic trajectories constructed from the offline dataset. In this paper, we introduce a novel conditional generative modeling approach to produce trajectories toward high-scoring regions. First, we construct synthetic trajectories toward high-scoring regions using the dataset while injecting locality bias for consistent improvement directions. Then, we train a conditional diffusion model to generate trajectories conditioned on their scores. Lastly, we sample multiple trajectories from the trained model with guidance to explore high-scoring regions beyond the dataset and select high-fidelity designs among generated trajectories with the proxy function. Extensive experiment results demonstrate that our method outperforms competitive baselines on Design-Bench and its practical variants. The code is publicly available in \texttt{https://github.com/dbsxodud-11/GTG}.