richert/pyrecu

Python library for recurrent neural network modeling and reservoir computing.

PyRecu

PyRecu is a Python library for recurrent neural network (RNN) modeling and reservoir computing (RC), developed by Richard Gast. It is very much a work-in-process kind of open-source project that everyone is welcome to contribute to.

Basic Features

1. rnn

• module for the forward simulation of RNN dynamics
• create the network data of your RC workflow
• runtime optimization via Numba, the behavior of which can be fully controlled by the user
• provide your own function implementing the integration of the RNN or use one of the provided RNNs in pyrecu.neural_models
• provide your own function decorators and function decorator arguments for runtime optimization

2. readout

• module for model fitting based on previously simulated network data
• train and test the readout layer of your RC workflow
• use ridge regression and cross-validation features

3. util

• collection of utility functions
• create connectivity matrices with random sampling from spatial kernels
• calculate the sequentiality of the network dynamics
• calculate the functional modularity of the network from simulated RNN dynamics

Installation

To install PyRecu, clone this repository and run the following line from the directory in which the repository was cloned:

python setup.py install

This will also install the requirements of the software listed below.

Dependencies

• numpy
• numba
• scikit-learn
• scipy

Contact

If you have any questions, want to contribute to the software, or just get in touch, feel free to post an issue or contact Richard Gast.