Invitation ~ Fundamentals of Active Inference 2023

Are you interested in gaining a deep understanding of active inference and the free energy principle (AIF-FEP) and learning to implement it from scratch in Python?

Join Fundamentals of Active Inference, a new Active Inference Institute project led by Sanjeev Namjoshi. Sanjeev is working on a hands-on tutorial-style textbook where the Active Inference framework and Free Energy Principle are explored from first principles by engaging with the literature, mathematics, and directly integrating with Juypter notebooks written in Python. To sign up please fill out this form.

Weekly meetings will start at 21 UTC on January 26th, 2023.

Meetings will continue through 2023 on Thursdays at 21 UTC. Meetings will be recorded, and there will be asynchronous work in terms of reading the chapters and preparing feedback.

The project goal is to get feedback on the work-in-progress textbook, covering a chapter every two weeks. For each chapter, the first week will include an overview of the chapter contents presented by Sanjeev. During this session, the math and contents will be discussed in detail and code examples accompanying the chapter will be shown. During the week that follows, participants should then read the draft of the chapter themselves and examine the accompanying Jupyter notebooks to provide feedback/corrections for the second week’s session. We value all levels of participation - from those that just want to read and give suggestions of how to make things clearer to those who would like to correct and improve the code and add clarity to the content.

The project and book’s contents are aimed at those that have basic familiarity with statistics, probability theory, calculus, and Python programming. However, we anticipate that the course may still be of interest to those that do not have familiarity in these areas. We value the input and feedback from all levels - from beginners to experts. We will start from simple simulations and build up to complex active inference models. We will also implement and discuss the details behind Bayesian mechanics from first principles, showing how Markov blankets and other constructs related to the Free Energy Principle can be implemented in Python. With the community’s participation and interaction we hope to gain valuable feedback in order to make this book as clear and well-written as possible in order to appeal to a large audience.

To sign up please fill out this form and you will receive more information about where to join and participate.

Let us know by reply if you have any ideas or questions.

Thank you for reading and for the consideration.

Active Inference Institute

Thanks for reading Active Inference Institute! Subscribe for free to receive new posts and support my work.