June 22, 2020 to July 2, 2020
US/Central timezone

Convolutional Neural Network Applied to BiPo Discrimination at SoLid

Not scheduled


Dr Victor Buridon (LPC Caen)


The BiPo-214 decay is a significant background to the antineutrino reconstruction in the SoLid detector. It has a very similar time and spatial signature as of the product of the inverse beta decay. SoLid has developed a unique technique to separate scintillation signals of neutron captured on Lithium-6 from the Po-214 alpha. The current Pulse Shape Discrimination technique is based on a ratio of charge integration window. Using a 1-dimensional convolutional network to fully characterise the shape of the waveform, we demonstrated a factor of two to three improvement in discrimination power over the current technique.


Deployment of a CNN reduces further the BiPo background in SoLid over the current PSD method.

Experiment/Collaboration SoLid

Primary author

Dr Victor Buridon (LPC Caen)


Dr Antonin Vacheret (Imperial College London) Dr Gregory Lehaut (LPC Caen)

Presentation materials