vande

Variational Autoencoding for Anomaly Detection

setup

requires Python3 and TF 2.X

Particle VAE

train

main_train_particle_vae.py

parameters

• run_n ... experiment number
• beta ... beta coefficient for Kullback-Leibler divergence term
• loss ... 3D+KL loss or MSE+KL loss (from losses module)
• reco_loss ... 3D or MSE (from losses module)
• cartesian ... True/False: constituents coordinates (if False: cylindrical)

predict

main_predict_particle_vae.py

parameters

• run_n ... experiment number
• cartesian ... True/False: constituents coordinates (if False: cylindrical)