Submission information
ActiveInference.jl is a Julia package for building and simulating intelligent agents based on the principles of active inference. This framework, inspired by neuroscience, models how living beings (and potentially, artificial ones) perceive, learn, and make decisions. The package provides the tools to create these models as partially observable Markov decision processes (POMDPs), allowing researchers and developers to simulate agent behavior and fit models to real-world data.
The primary audience for ActiveInference.jl includes researchers in computational psychiatry and developers working on cutting-edge artificial intelligence. Its applications span a wide range of AI domains. In essence, ActiveInference.jl provides the building blocks for creating AI that is not just programmed, but that actively seeks to understand and engage with its world in a way that is both intelligent and adaptable.