English

Relaxing the Constraints on Predictive Coding Models

Neurons and Cognition 2020-10-13 v2 Artificial Intelligence Machine Learning

Abstract

Predictive coding is an influential theory of cortical function which posits that the principal computation the brain performs, which underlies both perception and learning, is the minimization of prediction errors. While motivated by high-level notions of variational inference, detailed neurophysiological models of cortical microcircuits which can implements its computations have been developed. Moreover, under certain conditions, predictive coding has been shown to approximate the backpropagation of error algorithm, and thus provides a relatively biologically plausible credit-assignment mechanism for training deep networks. However, standard implementations of the algorithm still involve potentially neurally implausible features such as identical forward and backward weights, backward nonlinear derivatives, and 1-1 error unit connectivity. In this paper, we show that these features are not integral to the algorithm and can be removed either directly or through learning additional sets of parameters with Hebbian update rules without noticeable harm to learning performance. Our work thus relaxes current constraints on potential microcircuit designs and hopefully opens up new regions of the design-space for neuromorphic implementations of predictive coding.

Keywords

Cite

@article{arxiv.2010.01047,
  title  = {Relaxing the Constraints on Predictive Coding Models},
  author = {Beren Millidge and Alexander Tschantz and Anil Seth and Christopher L Buckley},
  journal= {arXiv preprint arXiv:2010.01047},
  year   = {2020}
}

Comments

02/10/20 initial upload; 10/10/20 minor fixes