A training manual for Bayesian mechanics

I went to see the University of Strathclyde’s Mathematically Structured Programming Group to speak about Bayesian mechanics and its possible relations to categorical cybernetics.

Abstract:

Bayesian mechanics is a new set of tools for studying the mathematical physics of coupled random dynamical systems. In this talk I will provide an overview of the key devices of Bayesian mechanics and demonstrate how they are used. I will focus on what makes Bayesian mechanics inferential; that is, what makes it a ‘physics of and by beliefs.’ Particular emphasis will be placed on the coupling structure underlying Bayesian mechanics, with an eye towards formalisation in terms resembling those of decorated cospans or categories of lenses (e.g., Bayesian open games).

Talk: link to recording (access password: 5kh.AgX?)

Slides: MSP July 2022.pdf