Physics-Informed Uncertainty Enables Reliable AI-driven Design

2026-01-26

AI Summary

This work addresses the critical limitation of conventional deep learning surrogate models in inverse design—namely, their inability to provide reliable uncertainty quantification, particularly in data-sparse regions where optimization often fails. To overcome this, the authors introduce a novel paradigm termed “physics-informed uncertainty,” which leverages the degree to which model predictions violate physical laws as a proxy for uncertainty, replacing traditional uncertainty estimation methods. This approach is integrated into a multi-fidelity, uncertainty-aware optimization framework tailored for the inverse design of frequency-selective surfaces operating in the 20–30 GHz band. The proposed method substantially enhances both reliability and efficiency in high-dimensional inverse design, increasing success rates from below 10% to over 50% while reducing computational cost by an order of magnitude, thereby demonstrating its efficacy and robustness in high-frequency electromagnetic structure design.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Reasoning under Uncertainty: Stochastic Optimization
• Search and Optimization: Mixed Discrete/Continuous Search

Application Category

• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Psychology-informed user models and recommender systems

Abstract

Inverse design is a central goal in much of science and engineering, including frequency-selective surfaces (FSS) that are critical to microelectronics for telecommunications and optical metamaterials. Traditional surrogate-assisted optimization methods using deep learning can accelerate the design process but do not usually incorporate uncertainty quantification, leading to poorer optimization performance due to erroneous predictions in data-sparse regions. Here, we introduce and validate a fundamentally different paradigm of Physics-Informed Uncertainty, where the degree to which a model's prediction violates fundamental physical laws serves as a computationally-cheap and effective proxy for predictive uncertainty. By integrating physics-informed uncertainty into a multi-fidelity uncertainty-aware optimization workflow to design complex frequency-selective surfaces within the 20 - 30 GHz range, we increase the success rate of finding performant solutions from less than 10% to over 50%, while simultaneously reducing computational cost by an order of magnitude compared to the sole use of a high-fidelity solver. These results highlight the necessity of incorporating uncertainty quantification in machine-learning-driven inverse design for high-dimensional problems, and establish physics-informed uncertainty as a viable alternative to quantifying uncertainty in surrogate models for physical systems, thereby setting the stage for autonomous scientific discovery systems that can efficiently and robustly explore and evaluate candidate designs.

Problem

• inverse design
• uncertainty quantification
• surrogate models
• physics-informed
• optimization

Innovation

• Physics-Informed Uncertainty
• Inverse Design
• Uncertainty Quantification
• Multi-fidelity Optimization
• Frequency-Selective Surfaces