Aug 10 – 14, 2020
Fermi National Accelerator Laboratory
US/Central timezone
All registrants were sent Zoom connection info by email on Sunday, August 9. Connection link emails repeated daily at 7 AM (CDT).

A deep-learning based raw waveform region-of-interest finder for the liquid argon time projection chamber

Aug 10, 2020, 1:10 PM


Wanwei Wu (Fermilab)


Deep learning techniques are being widely used in high energy physics and they are playing a significant role in the reconstruction of the neutrino interactions in particle detectors. However, those algorithms normally use 2D images as inputs. Here, we consider a unique approach of using a simple 1D convolutional neural network (1D-CNN) to look directly at raw waveforms from single wires in a Liquid Argon Time Projection Chamber (LArTPC). In this poster, we present encouraging results in the application of a 1D-CNN to the task of finding the region-of-interest (ROI) in raw LArTPC waveforms from data collected by the ArgoNeuT experiment. The 1D-CNN ROI finder shows a promising ability to extract small signals from low-energy phenomena and can be implemented in early stages of reconstruction as a very effective filter to remove noise. It offers great potential for low-energy neutrino physics.

Primary authors

Wanwei Wu (Fermilab) Lorenzo Uboldi (University of Milano) Michael H L Wang (Fermilab) Tingjun Yang (Fermilab)

Presentation materials