Torchhd: A Python Library for Hyperdimensional Computing and Vector Symbolic Architectures
Authors: Mike Heddes, Igor Nunes, Pere Vergés, Denis Kleyko, Danny Abraham, Tony Givargis, Alexandru Nicolau, Alexander Veidenbaum; Journal of Machine Learning Research, 24(255):1−10, 2023.
Abstract
Hyperdimensional computing (HD), also known as vector symbolic architectures (VSA), is a framework that utilizes random high-dimensional vector spaces for distributed representations. The progress and advancement of this multidisciplinary field heavily rely on the collaboration and dissemination of research within the scientific community. In this regard, we introduce Torchhd, an open source Python library designed to support HD/VSA research. Torchhd aims to enhance the accessibility of HD/VSA and provides a solid foundation for further research and application development. Built on top of PyTorch, this user-friendly library offers state-of-the-art HD/VSA functionality, well-documented resources, and implementation examples from reputable publications. Performance evaluations comparing Torchhd with publicly available code indicate that experiments can run up to 100 times faster. Torchhd can be accessed at: https://github.com/hyperdimensional-computing/torchhd.
[abs]
[code]