/learn_error_by_generalization

Experiment consisting in learning error using an explicit generalization loss; implemented in Tensorflow

Primary LanguagePythonMIT LicenseMIT

Learn error by explicit generalization (WIP)

Learn to generalize explicitly by learning error vector on testset batch

Run instructions

Assuming you installed Python 3 with tensorflow, scikit-learn and numpy you need to:

  • Clone the project
git clone https://github.com/PaulEmmanuelSotir/generalization_loss.git
cd ./explicit_generalization_training
  • Train the model on CIFAR dataset
python train_dnn.py
  • Apply explicit generalization training on trained model to learn errors on testset predictions
python generalization_training.py