Deriving time-averaged active inference from control principles

Autor:	Sennesh, Eli, Theriault, Jordan, van de Meent, Jan-Willem, Barrett, Lisa Feldman, Quigley, Karen
Rok vydání:	2022
Předmět:	Electrical Engineering and Systems Science - Systems and Control Quantitative Biology - Neurons and Cognition Statistics - Machine Learning
Druh dokumentu:	Working Paper
Popis:	Active inference offers a principled account of behavior as minimizing average sensory surprise over time. Applications of active inference to control problems have heretofore tended to focus on finite-horizon or discounted-surprise problems, despite deriving from the infinite-horizon, average-surprise imperative of the free-energy principle. Here we derive an infinite-horizon, average-surprise formulation of active inference from optimal control principles. Our formulation returns to the roots of active inference in neuroanatomy and neurophysiology, formally reconnecting active inference to optimal feedback control. Our formulation provides a unified objective functional for sensorimotor control and allows for reference states to vary over time. Comment: Camera ready for International Workshop on Active Inference (IWAI) 2022
Databáze:	arXiv
Externí odkaz:	http://arxiv.org/abs/2208.10601 Zobrazit plný text záznamu View this record from Arxiv