Christopher D. Stiles, Ph.D.1,2
1Research and Exploratory Development Department, Johns Hopkins University Applied Physics Laboratory
2Department of Mechanical Engineering, Johns Hopkins University
“Advancing Closed-Loop AI-Driven Materials Discovery”
Abstract: Artificial Intelligence and Machine learning (AI/ML) presents tremendous opportunities to accelerate materials design and discovery, but adapting approaches from other domains remains challenging. Materials data tends to be sparser and less homogeneous than in other fields, and optimizing multiple properties simultaneously further complicates modeling. We present a closed-loop system integrating AI/ML predictions with synthesis and characterization to generate new labeled data, progressively improving models. We first demonstrate discovering new superconductors, then expand to predict mechanical properties as well, enabling multi-property design. We then assess and enhance key steps in this closed loop discovery cycle. Physics-guided generative models produce synthesizable candidate materials, multi-property predictions computationally screen materials, and high-throughput virtual screening focuses experiments on promising candidates via uncertainty-sampling and active learning refines models by emphasizing scientifically interesting regions based on experimental feedback. This fusion of generative modeling, predictive screening, uncertainty-aware acquisition, and iterative model refinement efficiently navigates vast materials spaces by unifying computational and experimental insights. Through targeted AI/ML development and integration, targeted materials design and discovery will become commonplace.
BIO: Dr. Christopher Stiles is the Chief Scientist of the Electrical and Mechanical Engineering Group at the Johns Hopkins University Applied Physics Laboratory where he is responsible for strategic planning and execution of engineering and research program. Prior to his current appointment, he served as Supervisor of the Multiscale Mathematical Modeling Section focusing on novel computational methods and artificial intelligence tools to enable scientific discovery. Dr. Stiles has an extensive publication record and research interests in topics related to multiscale modeling, machine learning, high performance computing, and computational physics. A major focus of his is convergent research between data scientists and physical scientists specializing in fields ranging from quantum engineering, to materials science, to chemistry, to biology, and ISRU. He also holds appointments as an Assistant Research Professor in the Department of Mechanical Engineering and the Vice Chair of Mechanical Engineering for the Engineering for Professionals Program within the Johns Hopkins University’s Whiting School of Engineering. Dr. Stiles is also a Hopkins Extreme Materials Institute Fellow. He earned his Bachelor of Science degrees in Physics and Mathematics from the University at Albany as well as a PhD in Nanoscale Science and Engineering from the College of Nanoscale Science and Engineering.