Towards Automated Discovery: A Review of Generative Models, Multimodal Learning and Closed-Loop Workflows in Inverse Materials Design

2026-06-01

AI Summary

This work addresses the limitations of traditional materials discovery, which relies on forward prediction and struggles to efficiently generate novel structures meeting specific performance criteria and physical constraints. The study systematically reviews and integrates generative models—such as variational autoencoders and diffusion models—with multimodal representations that fuse structural, electronic, spectral, and textual data, alongside closed-loop optimization strategies including Bayesian optimization, reinforcement learning, and active learning. It proposes a unified inverse design framework explicitly targeting synthesizability and stability. For the first time, the paper introduces a standardized evaluation paradigm emphasizing candidate materials’ validity, novelty, uniqueness, stability, and cost, while identifying critical failure modes such as mode collapse and distributional shift, thereby advancing the development of high-fidelity, automated workflows for new materials discovery.

Abstract

Inverse materials design is shifting materials discovery from forward prediction to targeted proposal of candidates that satisfy objectives under physical constraints. Here, we review recent advances in generative crystal structure modeling, multimodal learning, and closed-loop design pipelines for crystalline solids. We survey how modern generators learn chemical-structural priors from large databases to enable controllable sampling of periodic structures, and compare leading model classes including variational autoencoders, normalizing flows, autoregressive formulations, and diffusion models. Particular attention is given to how feasibility constraints and physical priors are enforced across the workflow, through representation choices, training objectives, sampling-time guidance, and post-generation screening and relaxation. We also discuss how multimodal learning fuses diverse materials modalities, including crystal structures, thermodynamic, electronic information, microscopy, spectroscopy, processing context, and scientific text, to construct a more universal, transferable representation of chemical space. In addition, diverse inverse-design strategies are examined, particularly those that integrate conditional generation with latent optimization, Bayesian optimization, reinforcement learning, and active learning. Finally, we highlight recurring failure modes, such as surrogate exploitation, diversity collapse, distribution shift, and the stability-synthesizability gap, and outline discovery-grade evaluation practices based on staged reporting of validity, novelty, uniqueness, stability, and cost.

Problem

• inverse materials design
• generative models
• multimodal learning
• closed-loop workflows
• crystal structure generation

Innovation

• inverse materials design
• generative models
• multimodal learning
• closed-loop workflows
• crystal structure generation