Nowadays, for experiments on microscopic imaging and ultrafast dynamics, precise and complete detection of multihit signals is indispensable. Demand for the discrimination of multiple particles that are simultaneously detected is growing for scientific purposes, such as probing electron-electron correlations and Coulomb explosion imaging. The particles collected with nearly the same arrival times will generate overlapping pile-up signals, which may cause signal loss or misjudgment. Here, we propose a method employing convolutional neural networks to accurately discriminate pile-up signals. This method is demonstrated with experimental photoionization data and achieves nearly an order of magnitude reduction in channel dead time for delay-line detectors. Moreover, this method has the potential to expand the applications of other kinds of detectors in various scientific fields.

Close

PrevNext

View figure in article

• Metrology
• Single- and few-photon ionization & excitation
• Ultrafast phenomena
• Atoms
• Convolutional neural networks
• Deep learning
• Photoelectron techniques
• Photoionization