Estimation of belief states in superposition, after the world model has been trained.

<p>To minimize its prediction error due to faulty belief state estimation, an agent considers multiple clone clips as candidate belief states simultaneously. For the initial observation, (on the left), the agent includes all corresponding clone clips to its hypothesis, as depicted on the right...

Full description

Saved in:

Bibliographic Details
Main Author:	Joséphine Pazem (22184363) (author)
Other Authors:	Marius Krumm (22184366) (author), Alexander Q. Vining (11320591) (author), Lukas J. Fiderer (4865587) (author), Hans J. Briegel (6383642) (author)
Published:	2025
Subjects:	Science Policy Biological Sciences not elsewhere classified Information Systems not elsewhere classified using internal rewards partially observable grid free energy principle expected free energy agents &# 8217 timed response task various reinforcement learning partially observable environments feps agents build interpretability </ p build upon model agents navigation task given task feps ). feps model xlink "> understanding aspects term goals target observation results show recent work prediction accuracy observations based multidisciplinary interest mathematical models last decade including elements constraints imposed complex environments behavioral biology appropriately contextualizing active inference
Tags:	Add Tag No Tags, Be the first to tag this record!

Estimation of belief states in superposition, after the world model has been trained.

Similar Items