Machine Learning

Machine learning (ML) algorithms have wide application in LArTPC data processing, serving to filter raw data, simulate various event features, and reconstruct and classify events and the particles within them, among other uses. The following list highlights a few ML algorithms that address different reconstruction and classification problems and that have been integrated into LArSoft.

  1. NuGraph2 GNN event reconstruction
  2. CNN-based event classification and particle ID using CVN
  3. Waveform region-of-interest finding for supernova event triggering

In some cases, these algorithms have been written in the context of a particular experiment, but are easily generalizable to be detector agnostic.