The MDiscovery Challenge is a unique opportunity offered in conjunction with the TMS Specialty Congress 2025. This machine learning challenge is designed to propel materials and manufacturing discovery and research into the future. It aims to predict properties from process and/or structural data streams while identifying methodologies and machine-learning practices that yield the best predictions. This challenge is for university and government researchers, industry professionals, and students in the fields of materials and manufacturing. It provides a unique opportunity to harness the power of machine learning to predict properties from process and microstructural data streams, ultimately advancing materials and manufacturing innovations.
Challenge Overview
The purpose of this challenge is two-fold: to predict properties from process and/or microstructural data streams and to identify methodologies and machine-learning practices that yield the best predictions. The MDiscovery challenge provides participants with an exciting opportunity to tackle real-world problems in materials and manufacturing.
Challenge Tasks
Participants will predict various properties, including hardness, electrical resistivity, and fracture toughness, utilizing provided data for training and calibration. Bonus questions on solving inverse problems will also be available, fostering creativity and innovation in methodology. Results should be submitted in the requested format, including quantities of interest and calibration/training results for comparison. Specific details on the questions and judging rubric will be provided at a later date.
Data Sets
Our dedicated team is diligently working on cleaning up, verifying, and curating the data sets (using FAIR data standards). Data sets are set for release soon.
Together, let's unlock the potential of machine learning in materials and manufacturing. Stay tuned for further updates.