Hodgkin-Huxley Backpropogation

James Hazelden, Michael Ivanitskiy, and Daniel Forger

NeurIPS 2020 Workshop on Beyond Backpropagation: Novel Ideas for Training Neural Architectures.

Paper page PDF

Description

Classical artificial neural networks have been able to solve a variety of tasks in machine vision and machine learning. They take their inspiration from the ability of neural connectivity to represent a wide range of functions. In this paper, we explore the application of stochastic gradient descent (i.e. backpropogation) to networks of neurons with more complex internal behavior than in traditional artificial neural networks. In particular, we derive a backpropogation type algorithm for the feed-forward networks made up of the ubiquitous Hodgkin-Huxley neuronal model.

Citation

James Hazelden, Michael Ivanitskiy, and Daniel Forger. Hodgkin-Huxley Backpropogation. NeurIPS 2020 Workshop on Beyond Backpropagation: Novel Ideas for Training Neural Architectures, 2020.

BibTeX

@inproceedings{hazelden2020hodgkinhuxley,
  title={Hodgkin-Huxley Backpropogation},
  author={James Hazelden and Michael Ivanitskiy and Daniel Forger},
  year={2020},
  booktitle={NeurIPS 2020 Workshop on Beyond Backpropagation: Novel Ideas for Training Neural Architectures},
  note={Workshop poster},
  url={PDF_LINK_TO_ADD}
}