ProtoTorch: Prototype Learning in PyTorch

Tensorflow users, see: ProtoFlow

Description

This is a Python toolbox brewed at the Mittweida University of Applied Sciences in Germany for bleeding-edge research in Prototype-based Machine Learning methods and other interpretable models. The focus of ProtoTorch is ease-of-use, extensibility and speed.

Installation

ProtoTorch can be installed using pip.

pip install -U prototorch

To also install the extras, use

pip install -U prototorch[all]

Note: If you're using ZSH, the square brackets [ ] have to be escaped like so: \[\], making the install command pip install -U prototorch\[all\].

To install the bleeding-edge features and improvements:

git clone https://github.com/si-cim/prototorch.git
git checkout dev
cd prototorch
pip install -e .[all]

Documentation

The documentation is available at https://prototorch.readthedocs.io/en/latest/

Usage

For researchers

ProtoTorch is modular. It is very easy to use the modular pieces provided by ProtoTorch, like the layers, losses, callbacks and metrics to build your own prototype-based(instance-based) models. These pieces blend-in seamlessly with Keras allowing you to mix and match the modules from ProtoFlow with other modules in torch.nn.

For engineers

ProtoTorch comes prepackaged with many popular Learning Vector Quantization (LVQ)-like algorithms in a convenient API. If you would simply like to be able to use those algorithms to train large ML models on a GPU, ProtoTorch lets you do this without requiring a black-belt in high-performance Tensor computing.

Bibtex

If you would like to cite the package, please use this:

@misc{Ravichandran2020b,
  author = {Ravichandran, J},
  title = {ProtoTorch},
  year = {2020},
  publisher = {GitHub},
  journal = {GitHub repository},
  howpublished = {\url{https://github.com/si-cim/prototorch}}
}