{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/deep-active-inference","title":"Deep Active Inference","arxiv_id":"1709.02341","date":"2018-10-11","proceeding":null,"authors":[],"abstract":"This work combines the free energy principle from cognitive neuroscience and\nthe ensuing active inference dynamics with recent advances in variational\ninference in deep generative models, and evolution strategies to introduce the\n\"deep active inference\" agent. This agent minimises a variational free energy\nbound on the average surprise of its sensations, which is motivated by a\nhomeostatic argument. It does so by optimising the parameters of a generative\nlatent variable model of its sensory inputs, together with a variational\ndensity approximating the posterior distribution over the latent variables,\ngiven its observations, and by acting on its environment to actively sample\ninput that is likely under this generative model. The internal dynamics of the\nagent are implemented using deep and recurrent neural networks, as used in\nmachine learning, making the deep active inference agent a scalable and very\nflexible class of active inference agent. Using the mountain car problem, we\nshow how goal directed behaviour can be implemented by defining appropriate\npriors on the latent states in the agent's model. Furthermore, we show that the\ndeep active inference agent can learn a generative model of the environment,\nwhich can be sampled from to understand the agent's beliefs about the\nenvironment and its interaction therewith.","url_abs":"http://arxiv.org/abs/1709.02341v5","url_pdf":"http://arxiv.org/pdf/1709.02341v5.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"deep-active-inference","repo_url":"https://github.com/kaiu85/deepAI_paper","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"none","reach":{"status":"unanswered"}}],"tasks":[{"task_slug":"variational-inference","task_name":"Variational Inference"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1709.02341","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}