Related Experiment Video
Updated: Feb 11, 2026

Label-Free Identification of Lymphocyte Subtypes Using Three-Dimensional Quantitative Phase Imaging and Machine Learning
Published on: November 19, 2018
Machine learning-enabled identification of material phase transitions based on experimental data: Exploring
Linglong Li1,2,3, Yaodong Yang3, Dawei Zhang4
1Center for Nanophase Materials Sciences, Oak Ridge National Laboratory, Oak Ridge, TN 37831, USA.
Researchers developed a new machine learning method to identify phase transitions in ferroelectric materials. This approach analyzes piezoelectric relaxation data to construct temperature-bias phase diagrams without needing to measure the order parameter.
Area of Science:
- Condensed matter physics
- Materials science
Background:
- Phase transitions and phase diagrams are crucial in condensed matter physics and materials science.
- Experimental studies typically require comprehensive scattering, thermodynamics, and modeling.
Purpose of the Study:
- To present a novel data mining approach using machine learning to identify structural phase transitions.
- To construct temperature-bias phase diagrams for relaxor crystals without measuring the order parameter.
- To apply this method to nanometer-scale volumes probed by atomic force microscopy.
Main Methods:
- Utilizing piezoelectric relaxation studies to measure collective dynamics.
- Applying machine learning to analyze multidimensional datasets of relaxation responses to voltage and thermal stimuli.
- Validating the approach with simulations based on a two-dimensional Ising model.
Main Results:
- Successfully identified the onset of structural phase transitions in nanometer-scale volumes.
- Generated temperature-bias phase diagrams for a relaxor crystal.
- Demonstrated the method's suitability through simulations.
Conclusions:
- Machine learning offers a robust and statistically significant approach to determine phase transitions in ferroelectrics.
- This method provides a general way to identify critical regimes and phase boundaries.
- The technique bypasses the need for direct order parameter measurement.
Related Concept Videos
Phase Transitions
Phase Transitions: Sublimation and Deposition
Phase Transitions: Vaporization and Condensation
Phase Transitions: Melting and Freezing
Data Collection II
Data Collection I

