SpeckleNN: a unified embedding for real-time speckle pattern classification in X-ray single-particle imaging with limited labeled examples

Abstract

With X-ray free-electron lasers (XFELs), it is possible to determine the three-dimensional structure of noncrystalline nanoscale particles using X-ray single-particle imaging (SPI) techniques at room temperature. Classifying SPI scattering patterns, or `speckles', to extract single-hits that are needed for real-time vetoing and three-dimensional reconstruction poses a challenge for high-data-rate facilities like the European XFEL and LCLS-II-HE. Here, we introduce SpeckleNN, a unified embedding model for real-time speckle pattern classification with limited labeled examples that can scale linearly with dataset size. Trained with twin neural networks, SpeckleNN maps speckle patterns to a unified embedding vector space, where similarity is measured by Euclidean distance. We highlight its few-shot classification capability on new never-seen samples and its robust performance despite having only tens of labels per classification category even in the presence of substantial missing detector areas. Without the need for excessive manual labeling or even a full detector image, our classification method offers a great solution for real-time high-throughput SPI experiments.

Turning European XFEL raw data into user data

“…This is combined with expanding on the correction add-on mechanism to exploit data being present in the memory of high-performance GPUs to benefit custom user analysis as well. In fact, these devices are increasingly used to accelerate time-consuming analysis tasks in the analysis of X-ray experiments and enable their real-time application, in particular for techniques based on machine Frontiers in Physics frontiersin.org learning [43][44][45][46]. These capabilities are also the ideal place to apply data reduction.…”

Data reduction activities at European XFEL: early results

“…Additionally, most of the modern X-ray photon sources are exploring and developing data reduction strategies [22], also owing to initiatives of policymakers such as European Union's . Topics explored vary from lossless and lossy compression [31] to artificial intelligence [25,26,32,33], from dedicated hardware solutions to FAIR data [34].…”