| About this Abstract |
| Meeting |
2024 TMS Annual Meeting & Exhibition
|
| Symposium
|
Novel Strategies for Rapid Acquisition and Processing of Large Datasets from Advanced Characterization Techniques
|
| Presentation Title |
Using Video Games for Training Data on Microstructural Design |
| Author(s) |
Christopher W. Adair, Oliver K. Johnson |
| On-Site Speaker (Planned) |
Christopher W. Adair |
| Abstract Scope |
Searching the configuration space of Grain Boundary Networks (GBNs) for optimal design parameters is difficult due to the high dimensionality of the space (5N degrees of freedom). This naturally leads to wanting to implement machine learning (ML) for more effective searching of the space than stochastic methods. However, gaining training sets of structural changes leading to the property of interest in significant volume poses a challenge. In this talk we show how the publicly released video game "Operation: Forge the Deep" was able to gather a usable training set for optimizing a property on the GBN. We compare the performance of our agent based ML model against common stochastic methods and player performance. We compare the emergent character of GBNs generated by player outcomes and their similarities or differences. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Machine Learning, Computational Materials Science & Engineering, Modeling and Simulation |