search

Kladdy / neutrino-dnn

Deep-learning applied to neutrino property reconstruction
0 stars 0 forks source link

Study systematic uncertainties - 2020 dataset & EM shower #14

Closed Kladdy closed 3 years ago

Kladdy commented 3 years ago

Use my best network and try with Oscars dataset, both non-hadronic and hadronic.

Try first to just do predictions, then do training on data set

From Christian:

you can use this data set to test systematics from the askaryan emission model: /mnt/md0/data/SouthPole/single_surface_4LPDA_PA_15m_RNOG_fullsim.json/ARZ2020_emhad_noise.yaml/G03generate_events_full_surface_sim/LPDA_2of4_100Hz/4LPDA_1dipole_fullband/em_had_separately

use the files starting with data_had_emhad and the corresponding labels labels_had_emhad

the other files emhad_emhad contain the events with em showers

Kladdy commented 3 years ago

Ill make a 3x3 matrix as described in results part of report ;)