Physics-Informed Attention Mechanism and Generalization Capability of Deep Learning-Based Grain Growth Evolution Prediction

2026-06-15

AI Summary

This study addresses the limited out-of-distribution (OOD) generalization of deep learning models for grain growth prediction by introducing a grain boundary–oriented masked attention mechanism that embeds curvature-driven growth physics as an inductive bias into the network architecture. This design enables the model to automatically focus on grain boundary regions consistent with physical laws, significantly enhancing prediction robustness and accuracy under OOD conditions—such as real experimental microstructures, bimodal grain size distributions, and abnormal grain growth—without requiring retraining or fine-tuning. Experimental results demonstrate substantial improvements: on bimodal distribution tests, structural similarity (SSIM) increases from 0.6221 to 0.7609, and the average grain size prediction error decreases from 8.75% to 3.57%, confirming both the effectiveness and physical consistency of the proposed approach.

Abstract

Machine Learning (ML) models for grain growth prediction are typically trained on idealized synthetic data, yet practical applications require generalization to conditions outside the training distribution. This study evaluated the Out-Of-Distribution (OOD) generalization capability of the trained model from our previous study across three test cases, including experimental microstructures, microstructures characterized by a bimodal grain size distribution, and abnormal grain growth. To further probe whether physics-informed architectural design could improve robustness under these different conditions, a boundary-masked attention mechanism was proposed specifically for grain growth, constraining attention to grain boundary pixels. Both the baseline and the proposed physics-informed attention model were evaluated without retraining or fine-tuning on the OOD data. Both models successfully generalized to all three test cases, yet the boundary-masked attention mechanism provided substantial improvements, with the most notable gains for microstructures characterized by a bimodal grain size distribution, where Structural Similarity Index Measure (SSIM) improved from \num{0.6221} to \num{0.7609} and mean grain size ($\overline{R}$) error decreased from \SI{8.75}{\percent} to \SI{3.57}{\percent}. The attention heatmap analysis revealed that the boundary-masked attention model learned to concentrate attention on large grain boundaries in a manner consistent with curvature-driven grain growth physics, emerging from training without being explicitly encoded into the architecture. These results indicate that models trained on synthetic data can generalize to diverse OOD conditions without retraining, and that physics-informed attention may improve accuracy when the boundary morphology matches the training domain.

Problem

• grain growth
• out-of-distribution generalization
• physics-informed learning
• microstructure prediction
• deep learning

Innovation

• physics-informed attention
• grain growth prediction
• out-of-distribution generalization
• boundary-masked attention
• deep learning microstructure modeling