journal menu
issue contents
 | JOURNAL OF APPLIED CRYSTALLOGRAPHY |
ISSN: 1600-5767
Artificial neural networks: applications in X-ray photon science and crystallography
Edited by Tomas Ekeberg
This virtual collection gathers together some recent articles that describe applications of artificial intelligence to solving problems such as data classification, data analysis and parameter estimation in structural science. [A related virtual collection of articles in other IUCr journals is also available at https://journals.iucr.org/special_issues/2024/ML/.]
Cover illustration: Applying artificial neural networks to classification of neutron diffraction [Hao et al. (2023). J. Appl. Cryst. 56, 519–525], X-ray photon correlation spectroscopy [Timmermann et al. (2022). J. Appl. Cryst. 55, 751–757] and serial crystallography data [Nawaz et al. (2023). J. Appl. Cryst. 56, 1494–1504], as well as data preprocessing in single-particle imaging [Bellisario et al. (2022). J. Appl. Cryst. 55, 122–132].
Open 
access
access
Artificial intelligence is more present than ever, both in our society in general and in science. At the center of this development has been the concept of deep learning, the use of artificial neural networks that are many layers deep and can often reproduce human-like behavior much better than other machine-learning techniques. The articles in this collection are some recent examples of its application for X-ray photon science and crystallography that have been published in Journal of Applied Crystallography.
Open access
access
A convolutional neural network is applied for the single-hit diffraction-pattern classification step in single-particle-imaging experiments at X-ray free-electron lasers. This approach can be employed not only after the experiment but, importantly, also during an experiment and can significantly reduce the size of data storage for further analysis stages.
Open access
access
Two-time correlation maps are classified in a simulation framework using an auto-encoder network.
Open access
access
A machine learning method for distinguishing good and bad images in serial crystallography is presented. To reduce the computational cost, this uses the oriented FAST and rotated BRIEF feature extraction method from computer vision to detect image features, followed by a multilayer perceptron (neural network) to classify the images.
This work presents a machine-learning-assisted data reduction and analysis procedure for DEMAND, a single-crystal neutron diffractometer with a large-area position-sensitive detector at the High Flux Isotope Reactor, Oak Ridge National Laboratory, USA. This is a critical step to automate single-crystal neutron diffraction.
Open access
access
This article describes a qualitative study to unpack the internal workings of convolutional neural networks with the aim of visualizing information in the fundamental blocks of a standard network with serial crystallography data. Moreover, the study highlights region(s) or part(s) of an image that mostly contribute to a hit or miss prediction.
Open access
access
A number of machine-learning-based algorithms are presented for the reconstruction of gaps in experimental X-ray scattering images through inpainting approaches.
Open access
access
A neural network for coherent X-ray diffractive imaging experiments is presented that can restore noisy and masked simulated diffraction intensities from biological macromolecules.
The band gaps of metal–organic framework, organic and inorganic materials were assessed from X-ray powder diffraction patterns. The assessments were done with convolutional neural networks.
Open access
access
A new methodology is demonstrated for analyzing spatial temperature distributions using modeling and machine learning. The method is applied to X-ray data gathered during in situ laser melting.
Open access
access
A method is introduced to determine lattice parameters using machine learning. Analysis is presented of the impact of experimental conditions on machine learning prediction, and possibilities for automated unit-cell solution are explored.
Open access
access
An analysis approach for co-refinement of X-ray reflectivity measurement is presented, which works with sparsely sampled or noisy data for seven- to 200-fold speed increases.
Open access
access
A new artificial-intelligence-based platform, CrystalMELA, that can implement machine-learning models has been developed. Powder X-ray diffraction patterns of organic, inorganic and metal–organic compounds and minerals were used to train and test the learning models, and CrystalMELA has been employed for crystal system classification.
