Neuro-vector-symbolic architectures

Michael Hersche, Mustafa Zeqiri, Luca Benini, Abu Sebastian, Abbas Rahimi

Nature Machine Intelligence'23

This library implements the NVSA functionalities of the paper "A Neuro-vector-symbolic Architecture for Solving Raven's Progressive Matrices". A complete example for solving RAVEN can be found [here]

[Paper][Preprint]

Module Installation

This module can be installed using pip, and then simply used by import nvsa.

As a prerequisite you should have already installed torch and numpy.

From GitHub

To install module from GitHub use one of:

# Install using ssh
pip install [email protected]/IBM/neuro-vector-symbolic-architectures.git
# Install using https
pip install git+https://github.com/IBM/neuro-vector-symbolic-architectures.git

If you already have pytorch installed you might want to additionally use --no-dependencies flag:

# Install using ssh, without dependencies
pip install [email protected]/IBM/neuro-vector-symbolic-architectures.git --no-dependencies
# Install using ssh, without dependencies
pip install git+https://github.com/IBM/neuro-vector-symbolic-architectures.git --no-dependencies

From local repository

This version will automatically reflect any code changes that you make.

# Install from editable local version
pip install -e /path/to/local/neuro-vector-symbolic-architectures
# Also possible without dependencies
pip install -e /path/to/local/neuro-vector-symbolic-architectures --no-dependencies

Citation

If you use the work released here for your research, please cite the preprint of this paper:

@article{hersche2022neuro,
  title={A Neuro-vector-symbolic Architecture for Solving Raven's Progressive Matrices},
  author={Hersche, Michael and Zeqiri, Mustafa and Benini, Luca and Sebastian, Abu and Rahimi, Abbas},
  journal={Nat Mach Intell},
  year={2023}
}

License

Our code is licensed under Apache 2.0. Please refer to the LICENSE file for the licensing of our code.