Approximate Bayesian inference as a gauge theory

Abstract

In a published paper [Sengupta, 2016], we have proposed that the brain (and other self-organized biological and artificial systems) can be characterized via the mathematical apparatus of a gauge theory. The picture that emerges from this approach suggests that any biological system (from a neuron to an organism) can be cast as resolving uncertainty about its external milieu, either by changing its internal states or its relationship to the environment. Using formal arguments, we have shown that a gauge theory for neuronal dynamics -- based on approximate Bayesian inference -- has the potential to shed new light on phenomena that have thus far eluded a formal description, such as attention and the link between action and perception. Here, we describe the technical apparatus that enables such a variational inference on manifolds. Particularly, the novel contribution of this paper is an algorithm that utlizes a Schild's ladder for parallel transport of sufficient statistics (means, covariances, etc.) on a statistical manifold.

Extended version published in PLoS Biology, ICML 2017 Computational Biology Workshop (spotlight presentation)

References in corpus (2)

• Approximate Bayesian inference as a gauge theory
• Fast Variational Inference in the Conjugate Exponential Family

Cited by in corpus (9)

• Active inference on discrete state-spaces: a synthesis
• Approximate Bayesian inference as a gauge theory
• On Bayesian Mechanics: A Physics of and by Beliefs
• Path integrals, particular kinds, and strange things
• Is the brain relativistic?
• Bayesian mechanics of perceptual inference and motor control in the brain
• How Robust are Deep Neural Networks?
• Sentient Self-Organization: Minimal dynamics and circular causality
• Towards the neural code endowed in cortical microcolumns