June 2024 Newsletter 📖✍🌐 Active Inference Institute
Welcome to the June 2024 newsletter from the Active Inference Institute.
We are excited to bring many updates this month, at the scale of organizational developments as well as Research and Education activities.
For a up-to-date overview and presentation of all updates from the Institute, see our 2024 Quarterly Roundtable #2 video from today.
Institute scale
We are developing Partnerships with several organizations which are creatively applying Active Inference and looking to contribute to our open source ecosystem. If exploring a Partnership might be relevant for your organization or you know someone for whom this might be, we encourage you to get in touch or share information about this program.
The Institute is a 501(c)(3) tax exempt nonprofit organization by the United States IRS (donate here).
The rest of this newsletter is composed of our bundled updates from projects and people in the Active Inference ecosystem. Please complete a Measurement form if you have any updates to possibly include in a future month!
EduActive unit
The Textbook Group (Parr, Pezzulo, Friston 2022) project reports: We are preparing to begin Cohort 7 and continue Cohort 6, starting on July 15th. There are several exciting developments and differences about this upcoming interval of activity, including additional informal Math Group sessions, and RxInfer.jl usage for better understanding and applicability of textbook materials. Registration is open here.
The Fundamentals of Active Inference project reports: Working with Sanjeev Namjoshi and a group of committed learners, we completed working through the draft of his upcoming textbook. We eagerly await publication of Sanjeev’s work, which will provide a transformative improvement to the landscape of learning and applying Active Inference. This project will eventually reactivate at the Institute, for now we acknowledge and applaud the work of Sanjeev, and the participants who co-created this unique experience going back to the beginning of 2023.
The Audio-Visual Production project reports: Several great livestreams over the last month have included Livestream #057.2 with Thomas Parr “Active Data Sampling”, GuestStream #082.2 with Robert Worden on “Three-dimensional Spatial Cognition: Bees and Bats”, GuestStream #084.1 with Jesse G on “The Nature of Habits and Agential Systems”, and Active Inference Insights 023 with Iain McGilchrist.
From the Unifysica project, Ana Magdalena Hurtado reports: their group had weekly meetings (with Dean Tickles, Jeff Schulman, Hector Manrique) November 2023-May 2024. They wrote two commentaries, for Current Anthropology and the Physics of Life Reviews, and a National Institutes of Health Grant to fund NIH-led fellowship to work on social sciences ontologies for health and disease.
From the Action Research on Collective Foraging (Negotiation Affordances) project, Susan Hasty reports: Engaging and learning from our community has been both informative and fun. Meeting the wonderful members of our community, I've been truly amazed by the depth and breadth of insights and intelligence each and everyone has brought "to the table". It's been an honor to hear about the progress they’re making on their goals. I'm looking forward to seeing what comes next on our journey together as new opportunities and challenges arise. Thank you again for affording me this space to apply Active Inference to everyday negotiations. Also Susan has been accepted to an upcoming FA Summer Program on AI in Negotiation. She hopes to leverage the capabilities of our network in exploring how new A.I tools are being designed and built to support Humanitarian Negotiations (Crisis Management, Mitigation, Population Displacement, Mass Violence, and Social Unrest). If you have an interest in crafting AI-assisted negotiation tools comparing Reinforcement Learning with Active Inference, please reach out to her.
PabloFM reports from the Numinia project development along several lines releated to designing games that use Active Inference and also help learn it. The Numinia project by Numen Games develops games for community engagement that can help to work better. Some of the updates can be seen in the escape room and v0.1.0 brand and culture document.
From the MathArt project, Bert Berkers reports: We saw some applications of cognitive accounting in MathArt and look forward to exploring the imaginal expeditions of many others.
From the Active Inference Journal project, Alex Vyatkin reports: with Holly and Dave, we built a set of utilities in Python and Rust to transcribe YouTube videos using a local Whisper model. Used the WhisperX repository for segmentation and diarization, with models provided by Hugging Face. Tracked WAV conversion and transcription in SurrealDB.
ReInference unit
See here for all research & development work at the ReInference organizational unit.
The RxInfer.jl learning and development group project reports from the #engineering channel in Discord: We have made excellent progress in weekly meetings and development efforts between meetings. Multiple participants are contributing directly to the open source RxInfer.jl package which is exciting to see. The portfolio of RxInfer models made by Kobus Esterhuysen at LearnableLoop is a unique resource for people who are exploring the cutting edge of applying Active Inference. Great work everyone & we look forward to continued activity in the coming weeks as well.
The Active Inference Ontology project reports: Dave Douglass has led the production of two open access ontological works: “Aligning Active Inference Ontology to SUMO” and “Aligning Spatial Web Terms to SUMO”. These far-sighted works are building connections among several different ontology systems, including the core Active Inference ontology. As a group, several participants are exploring ontologies of and for mathematics. See here to learn more about the Ontology project.
Andrew Pashea from the project “Active Agents: An Active Inference Approach to Agent-Based Modeling in the Social Sciences” reports that Code and presentation slides are in progress and will be released as open source deliverables. The tutorial has been included in the program for the upcoming IC2S2 conference.
From the Symbolic cognitive robotics project, Jean-Francois Cloutier, an Institute Research Fellow reports that he is regularly posting on the #sym-cog-robotics Discord channel. See more updates at the project Github repo.
Feedback & Engagement
Please send in a Measurement if you have any Active Inference-related updates to share.
See all ongoing activities and projects, and join the Institute Discord, to get involved with our community.
Let us know via email to blanket@activeinference.institute if you have any other ideas or questions.
AII Officers