Explainable machine learning for diffraction patterns

Serial crystallography experiments at X-ray free-electron laser facilities produce massive amounts of data but only a fraction these are useful for downstream analysis. Thus, it is essential to differentiate between acceptable and unacceptable data, generally known as `hit' `miss', respectively. Image classification methods from artificial intelligence, or more specifically convolutional neural networks (CNNs), classify the into hit miss categories in order achieve reduction. The quantitative performance established previous work indicates that CNNs successfully serial desired [Ke, Brewster, Yu, Ushizima, Yang & Sauter (2018). J. Synchrotron Rad. 25 , 655–670], no qualitative evidence on internal workings has been provided. For example, there visualization highlight features contributing specific prediction while classifying experiments. Therefore, existing deep learning methods, including like `black box'. To this end, presented here study unpack with aim visualizing information fundamental blocks standard network data. region(s) part(s) an image mostly contribute visualized.