| Scope |
Artificial intelligence, machine learning, and data informatics (AI/ML/DI) are rapidly transforming the way materials are discovered, designed, and evaluated. These approaches create new opportunities to accelerate tasks such as screening vast candidate material spaces, extracting knowledge from heterogeneous data sources, guiding experiments and simulations, and developing predictive or generative models on timescales far shorter than traditional approaches alone. At the same time, broader adoption in materials science depends on addressing important challenges related to limited data availability, variable data quality, multimodal information, uncertainty, interpretability, and integration with established scientific knowledge and practice.
This symposium will focus on the frontiers of AI/ML/DI for materials discovery and development, with emphasis on connecting experiment, theory, and modeling in the modern materials research workflow. A central theme is the data-centric challenge of materials science: unlike many other fields, materials research is often constrained by sparse, noisy, heterogeneous, and difficult-to-integrate datasets spanning experiments, simulations, literature, and legacy knowledge sources. We therefore especially welcome contributions that address the generation, extraction, curation, management, and effective use of high-quality materials data, as well as methods that combine domain expertise and physics-based understanding with modern AI/ML approaches.
The symposium seeks to bring together researchers developing and applying AI/ML/DI in ways that strengthen the linkage between physical experimentation, computational modeling, and scientific decision-making. Contributions are invited on both fundamental advances and practical implementations that improve the efficiency, reliability, and impact of materials discovery.
Topics of interest include, but are not limited to:
· Uncertainty quantification, verification and validation, interpretability, and trust in AI/ML/DI for materials science
· Hybrid approaches that combine AI/ML/DI with physics-based models, mechanistic understanding, and established materials knowledge
· Large language models, multimodal foundation models, and agentic AI for materials property prediction, inverse design, service-life estimation, literature mining, and research workflow assistance
· Generative, scientific, and physics-informed machine learning for scarce, sparse, and multimodal materials datasets
· Novel approaches for AI-driven data generation, extraction, cleaning, annotation, fusion, and curation
· FAIR data principles, metadata frameworks, provenance, and interoperable infrastructure for materials data informatics
· Transfer learning, few-shot learning, federated learning, and related methods for data-limited materials problems
· Integration of experimental, computational, and literature-derived data to support materials discovery and decision-making
· AI-enabled discovery of new materials, compositions, processing-aware trends, constitutive relationships, and performance insights
· Methods and case studies that connect laboratory practice, computational experimentation, and AI-assisted materials development |