Scientific Achievement
An active learning approach based on co-navigation of the hypothesis and experimental spaces is developed.
Significance and Impact
This approach, actively testing alternative hypotheses during an experiment, can extend to higher-dimensional parameter spaces and more complex physical problems once the experimental workflow and hypothesis-generation mechanism are available.
Research Details
- Hypothesis-learning is realized via a combination of fully Bayesian structured Gaussian processes and reinforcement learning policy refinement
- Hypothesis-learning with Piezoresponse Force Microscopy is used for exploring concentration-induced phase transitions in combinatorial libraries of Sm-doped BiFeO3.
M. Ziatdinov, Y. Liu, A.N. Morozovska, E.A. Eliseev, X. Zhang, I. Takeuchi, and S.V. Kalinin, Advanced Materials (2022). DOI: 10.1002/adma.202201345