March 10, 2025
The aiM Program (AI for Understanding and Designing Materials) at Duke University supports student research across various fields bridging machine learning and materials science including contributions in decision-making algorithms and 3D printing for medical applications. Below are some of the recent contributions from students in the program.
Io Saito: Materials Simulation in the Brinson Group
Io Saito, a student working with the Brinson group led by Professor L. C. Brinson, co-authored "Pushing AFM to the Boundaries: Interphase Mechanical Property Measurements near a Rigid Body," published in Macromolecules. She also created the cover art for the journal issue. The research focuses on measuring mechanical properties at material interfaces, which is important for the development of efficient and durable composites. (https://doi.org/10.1021/acs.macromol.4c01993)
Yiyang Sun: Machine Learning in the Rudin Group
Yiyang Sun, a student working with the Rudin group, contributed to "Improving Decision Sparsity," presented at the Thirty-eighth Annual Conference on Neural Information Processing Systems. The study explores methods for optimizing machine learning models by reducing complexity while maintaining accuracy. (https://doi.org/10.48550/arXiv.2410.20483)
Yiyang has also published “Dimension Reduction with Local Adjusted Graphs” presented at the 39th Annual AAAI Conference on Artificial Intelligence. The study designs a new method for identifying and separating real clusters within the data that other DR methods may overlook or combine. (https://doi.org/10.48550/arXiv.2412.15426)
Jake Peloquin: Hierarchical Materials Research in the Gall Group
Jake Peloquin, working with the Gall group, utilizes machine learning to accelerate the mechanical characterization of hierarchical materials. He led development of performance prediction algorithms for these materials in “Prediction of Tensile Performance for 3D Printed Photopolymer Gyroid Lattices,” published in Materials & Design (https://doi.org/10.1016/j.matdes.2023.112126) and in "Printability and Mechanical Behavior as a Function of Base Material," published in Additive Manufacturing (https://doi.org/10.1016/j.addma.2023.103892). In addition, he co-authored the publication “Predicting Compressive Stress-Strain Behavior of Elasto-Plastic Porous Media via Morphology-Informed Neural Networks”, which has been accepted in Nature Communications Engineering.
Peloquin also co-authored the study "Structure-Performance Relationships of Multi-Material Jetting Polymeric Composites Designed at the Voxel Scale," published in Journal of Manufacturing Processes, which investigates how material distribution influences the performance of 3D-printed composites (https://doi.org/10.1016/j.jmapro.2024.10.009).
Darryl Taylor: Bioengineering in the Jones Group
Darryl Taylor, a student working with the Jones group, co-authored "Metal-Modified Nanocellulose as a Synthetic Biofilm Substitute for Mechanical Removal Testing," submitted to Nature Materials. The research investigates the use of nanocellulose for biofilm removal in medical and industrial applications.
Ariana Quek: Machine Learning and Molecular Dynamics in the Guilleminot Group
Ariana Quek, a student in the Guilleminot group, authored the paper "Enhancing Robustness in Machine-Learning-Accelerated Molecular Dynamics: A Multi-Model Nonparametric Probabilistic Approach," published in Mechanics of Materials. This research introduces a probabilistic framework to improve the reliability in high-energy molecular dynamics simulations. (https://doi.org/10.1016/j.mechmat.2024.105237)
Quek also published "Approximating Fracture Paths in Random Heterogeneous Materials: A Probabilistic Learning Perspective," which presents a generative modeling approach to approximate fracture paths and provide uncertainty estimates for fractures in random heterogeneous materials. (https://doi.org/10.1061/JENMDT.EMENG-7617)