About this Abstract |
Meeting |
2025 TMS Annual Meeting & Exhibition
|
Symposium
|
AI/Data Informatics: Computational Model Development, Verification, Validation, and Uncertainty Quantification
|
Presentation Title |
Enhancing Materials Discovery in Complex Composition Spaces: FUSE Meets Generative ML |
Author(s) |
Hasan Muhammad Sayeed, Christopher Collins, Taylor Sparks, Matthew Rosseinsky |
On-Site Speaker (Planned) |
Hasan Muhammad Sayeed |
Abstract Scope |
To explore vast composition spaces and predict novel crystal structures more efficiently, we are extending the capabilities of the Flexible Unit Structure Engine (FUSE) by integrating generative machine learning (ML) models. FUSE is a powerful tool widely used in computationally led materials discovery. Our new approach leverages generative ML models to predict crystal structures, moving beyond traditional methods that were constrained within known materials. In addition to generative models, we will incorporate conditional crystal structure generation functionality, enabling researchers to generate materials with desired properties. By integrating this advanced ML technique with the DFT-based FUSE, we aim to uncover novel compounds with optimized properties and significantly accelerate the materials discovery process. This generative and conditional approach offers the potential to rapidly develop targeted materials, opening up new possibilities in complex composition spaces. This project represents a promising step forward in the quest for novel materials with unprecedented functionalities. |
Proceedings Inclusion? |
Planned: |
Keywords |
Computational Materials Science & Engineering, Machine Learning, |