Skip to content

ekneale/watchman-ml

Repository files navigation

Machine learning projects for WATCHMAN:

vertex_reconstruction:

Performs reconstruction of interaction vertex for pre-selected pairs of events using keras.

  1. get_features_pair.cc extracts hit features for pairs of events from raw rat simulation data
  2. vertex_reconstruction_keras_optimise_hyperas.py performs simultaneous optimisation of hyperparameters with hyperas
  3. vertex_reconstruction_keras_optimise_gridsearch.py performs optimisation of individual hyperparameters
  4. vertex_reconstruction_keras_train.py constructs a model from the training data
  5. vertex_reconstruction_keras_predict.py makes predictions for the interaction vertex in x, y and z

Additional requirement: ROOT for vertex_reconstruction_keras_predict.py

To view the training and validation statistics graphically, run: tensorboard --logdir logs/train after the training stage and open the link generated.

About

Machine learning algorithms for WATCHMAN

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages