Add data augmentation ability
Add the ability for data augmentation to SaltDataset, configurable through the config file. Two potential uses for this are training with the inclusion of systematic variations for auto-calibration or introducing noise to training data to prevent overtraining.
The transforms are passed as Callables, meaning that they can be defined in the configuration file using the class_path
and init_args
arguments.
The GaussianNoise transform class is included which allows for adding noise with a desired mean and standard deviation to selected inputs on being loaded by the SaltDataset.
Review checklist:
-
CI Passing -
Comments addressed -
Source branch is up to date with target
Edited by Nicholas Luongo