Handy package for training output constrained neural networks. As of now, this package only contains constrained extreme learning machines (CELM), constrained multi layer neural networks will be added in the future.
The source code is contained in the constrained_learning folder. The implementation of the (C)ELM can be found in constrained_learning/learner.py.
To install the package, run the following commands inside this repository:
conda create -n celm python=3.10
conda activate celm
pip install -e .
To run the synthetic examples:
python constrained_learning\examples\flat_gaussian.py
python constrained_learning\examples\monotone_quadratic.py
python constrained_learning\examples\ode.py
MIT license - See LICENSE.