SGD method for entropy error function with smoothing L0 regularization for neural networks

Install Dependencies

pip install -r requirements.txt

Run Training

We need to prepare a config.yaml file in order to run a training job. An example of config.yaml file can be seen at ./config.yaml in this repository.

Next, we can start the training by the below command

python train.py --config <path-to-the-config-file>

Note: All learning curve figures and tables in the paper are based on the training logs which can be seen through this link.

Support

If you encounter any issue while using this repository, please open a new issue in the issue panel or contact me via email.