Image-based machine learning for materials science

Abstract

Materials research studies are dealing with a large number of images, which can now be facilitated via image-based machine learning techniques. In this article, we review recent progress of machine learning-driven image recognition and analysis for the materials and chemical domains. First, the image-based machine learning that facilitates the property prediction of chemicals or materials is discussed. Second, the analysis of nanoscale images including those from a scanning electron microscope and a transmission electron microscope is discussed, which is followed by the discussion about the identification of molecular structures via image recognition. Subsequently, the image-based machine learning works to identify and classify various practical materials such as metal, ceramics, and polymers are provided, and the image recognition for a range of real-scenario device applications such as solar cells is provided in detail. Finally, suggestions and future outlook for image-based machine learning for classification and prediction tasks in the materials and chemical science are presented. This article highlights the importance of the integration of the image-based machine learning method into materials and chemical science and calls for a large-scale deployment of image-based machine learning methods for prediction and classification of images in materials and chemical science.

Advancements in High‐Throughput Screening and Machine Learning Design for 2D Ferromagnetism: A Comprehensive Review

“…Currently, data in materials science can be broadly classified into four types: experimental and simulated material properties (such as physical, chemical, structural, thermodynamic, and dynamic properties), [64] chemical reaction data (including reaction rates, reaction temperatures, etc. ), [65] image data (such as scanning electron microscope images of materials and photos of material surfaces), [66] and data from literature sources. These data can be discrete (e.g., texts), continuous (e.g., vectors and tensors), or in the form of weighted graphs.…”

Machine learning classification of polar sub-phases in liquid crystal MHPOBC

“…They have shown particular success in medical imaging applications 4–6 especially in cancer research 7,8 for example in mitosis detection in breast cancer histology images. 9 The use of machine learning can currently be followed through all fields of sciences, for example biology through the use in biosensors, 10 in chemistry for drug discovery 11 and in the computer-aided synthesis planning, 12 in material science 13 and physics 14 for example in nano-photonics, 15 but also astronomy 16–18 and particle physics. 19…”

Pair-Variational Autoencoders for Linking and Cross-Reconstruction of Characterization Data from Complementary Structural Characterization Techniques

“…Machine learning and artificial intelligence are becoming invaluable methods that enable faster discovery, innovation, and automation in chemical and materials sciences and engineering. − In particular, machine learning has found tremendous use in automating materials’ structural analysis, which is a vital step in establishing the design–structure–property relationship of a novel material. − For example, structural characterization of materials often depends on microscopy imaging techniques (e.g., scanning electron microscopy (SEM), transmission electron microscopy (TEM), or atomic force microscopy (AFM)) to visualize nanoscale or microscale structural features or patterns . Deep learning models used for pattern recognition and image analysis have been adopted as a viable means to automatically extract structural information from microscopy images regarding the ordered arrangements of the molecules, types of ordered assembly, and detection of objects’ (e.g., nanoparticles, assembled domains) shapes and sizes. − There are unique challenges, however, with training machine learning models that are used for everyday photographic image analysis to analyze materials’ microscopy images. For example, compared to photographic images of everyday objects that often are easily recognizable, materials’ microscopy images require detailed metadata of the material, chemistry, synthesis conditions, and imaging process conditions as labels.…”