Cross-material inverse learning and knowledge transfer for property-driven design of advanced ceramics

In the rapidly advancing field of materials science, artificial intelligence (AI) has emerged as a powerful tool for enhancing material property prediction, inverse design, and process optimization, particularly for complex ceramic systems. This study proposes a unified AI-driven framework that integrates decision-tree-based imputation, generative data augmentation, and advanced learning paradigms for accurate component and process recommendation. Initially, CART-based imputation is employed to robustly handle missing data, establishing a reliable baseline across KNN, BaTiO 3 , and PZT ceramic datasets. To further enhance data diversity and generalization, Variational Autoencoder (VAE) augmentation is applied, resulting in consistent performance gains across all materials. Building upon this enriched data, a Transformer-based inverse learning model is developed to infer material components and processing parameters from target properties. Extensive experiments demonstrate systematic performance improvements at each stage of the framework. The proposed approach achieves R 2 scores of 0.930, 0.959, and 0.928 after augmentation for KNN, BaTiO 3 , and PZT, respectively, while multi-material joint learning further improves predictive accuracy to an R 2 score of 0.987. Moreover, cross-material transfer learning enables effective knowledge adaptation between ceramic systems, achieving R 2 scores up to 0.986, demonstrating strong generalization in data-scarce scenarios. Inverse design evaluations further confirm low reconstruction errors, validating the reliability of the proposed framework. Overall, the results highlight the effectiveness of combining structured preprocessing, generative augmentation, joint learning, and transfer learning within a Transformer-based inverse modeling pipeline. This work underscores the potential of advanced AI techniques to accelerate data-driven materials design and provides a scalable foundation for future intelligent materials engineering applications.