About this Abstract |
| Meeting |
2026 TMS Annual Meeting & Exhibition
|
| Symposium
|
Theory and Design of Metallic Glasses
|
| Presentation Title |
Can Machine Learning Predict Metallic Glasses? |
| Author(s) |
Yifei He, Jan Schroers |
| On-Site Speaker (Planned) |
Yifei He |
| Abstract Scope |
Glass formation is a complex process depending on thermodynamic and kinetic processes that are predominately controlled by the mixing of elements. As it exists in a vast composition space, it has been often considered as the quintessential materials science problem to be addressed with machine learning strategies. Surprisingly the success has been quite moderate. We quantify the ability of ML strategies to predict glass forming ability by comparing it to various benchmark models and identify two major challenges. One challenge is representing the relevant characteristics of an alloy that determines glass forming ability through appropriate features. The other fundamental challenge is the discreteness of atoms properties. To reduce such limitations, one must develop more accurate descriptions of the complex problem, often related to its mixing behavior to achieve better feature representations. To mitigate problems that arise from atomic discreteness, reducing the problem through subgrouping into similar groups is one strategy. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Machine Learning, Phase Transformations, |