Symmetry-aware recursive image similarity exploration for materials microscopy

Abstract

In pursuit of scientific discovery, vast collections of unstructured structural and functional images are acquired; however, only an infinitesimally small fraction of this data is rigorously analyzed, with an even smaller fraction ever being published. One method to accelerate scientific discovery is to extract more insight from costly scientific experiments already conducted. Unfortunately, data from scientific experiments tend only to be accessible by the originator who knows the experiments and directives. Moreover, there are no robust methods to search unstructured databases of images to deduce correlations and insight. Here, we develop a machine learning approach to create image similarity projections to search unstructured image databases. To improve these projections, we develop and train a model to include symmetry-aware features. As an exemplar, we use a set of 25,133 piezoresponse force microscopy images collected on diverse materials systems over five years. We demonstrate how this tool can be used for interactive recursive image searching and exploration, highlighting structural similarities at various length scales. This tool justifies continued investment in federated scientific databases with standardized metadata schemas where the combination of filtering and recursive interactive searching can uncover synthesis-structure-property relations. We provide a customizable open-source package (https://github.com/m3-learning/Recursive_Symmetry_Aware_Materials_Microstructure_Explorer) of this interactive tool for researchers to use with their data.

Accurate vertical nanoelectromechanical measurements

“…Some recent examples include Bayesian optimization, 70 unsupervised learning control and structure of domain walls, 71 defects in lead zirconate titanate (PZT), 72 and a synthesisstructure-property relationship discovery tool based on a large (roughly 25 k) PFM image database. 73 It is well known that models trained on these datasets will typically have microscope calibration and reproducibility limitations that stem from instrumental crosstalk, sample and probe state variations, and limited data sizes-all in addition to the sample properties and functionality questions that were presumably the motivation for the measurements in the first place. To enable these exciting new capabilities on a wider scope and to avoid "garbage-in, garbage-out" scenarios, it is important for the measurements to become as accurate and reproducible as possible.…”

Influence of the Symmetry Neural Network Morphology on the Mine Detection Metric

“…The use of symmetry in the morphology of a neural network is an effective way to improve its ability in order to recognise patterns and detect patterns in image databases [9,10].…”

Probe microscopy is all you need *

“…The first applications of modern DL techniques to scanning probe [175] and scanning transmission electron [53] microscopy were demonstrated in 2017 for the off-line data analysis. Since then, there has been a growing interest in using deep and ML to automate routine analysis of microscopy data (such as atom/defect/particle finding) [8,155,[176][177][178][179], learning symmetries [180], and to extract physically meaningful latent parameters explaining high-dimensional observations [138,143,144,157,[181][182][183]. More recently, the pre-trained DL models were used to identify on-the-fly objects of interests, such as domain walls and atomic defects, in automated experiments in scanning probe and electron microscopy.…”