To be honest, when I talk about the “free energy principle”, I typically have in mind a certain class of algorithmic implementations of it, involving generative models and using maximum likelihood estimation through online gradient descent to minimize their prediction errors. Something like https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1000211
To be honest, when I talk about the “free energy principle”, I typically have in mind a certain class of algorithmic implementations of it, involving generative models and using maximum likelihood estimation through online gradient descent to minimize their prediction errors. Something like https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1000211