Trying to apply the free energy principle to engineering problems, where 'trying' means: minimizing free energy.
Eindhoven, Netherlands
Joined January 2014
- Tweets213
- Following692
- Followers522
- Likes500
Bert de Vries retweeted
Out now in TMLR! 🚀
We formalize how epistemic priors transform Expected Free Energy minimization into a standard variational objective.
This allows us to frame planning as a continuous variational optimization problem, moving away from combinatorial tree search. 👇
I recently had the pleasure to lecture at the Machine Learning Summer School in Melbourne t.ly/hX0Uo on Bayesian Machine Learning → Active Inference.
All materials (slides + notebooks) available at github.com/bertdv/mlss-2026 . Thread below 👇
Finally: Active Inference (AIF). AIF extends BML to embodied agents with a full commitment to variational inference for state estimation, learning, planning, and control. See t.ly/zcSdH
Bert de Vries retweeted
This is what we ask ourselves in our latest paper (arxiv.org/abs/2508.02197), where we solve Minigrid environments with Active Inference. (Spoiler, it works!)
Code available at github.com/biaslab/EFEasVFE
Done together with @mlmykola , @ThijsvdLaar and @bertdv0
Bert de Vries retweeted
What's even nicer: because our method injects priors locally, everything still works within @ReactiveBayes ' RxInfer.jl using message passing. Special thanks to my colleagues at @LazyDynamics for making this possible!
Bert de Vries retweeted
In Active Inference, a lot of time is spent on computing Expected Free Energy. What if we could tweak the generative model such that EFE can be minimised with traditional variational inference methods?
Agreed with @fchollet on FEP (t.ly/s4gMV), but FEP is more than a pretty good idea, and there are more benefits to realizing an agent as an active inference (AIF) process beyond active data selection. I will mention a few below:
(6) Finally, FEP is more than a pretty good idea as it can be derived from first principles by information theory, see e.g., blog at t.ly/Bl2DO plus refs. An AIF process avoids ad hoc design choices often found in man-made AI algorithms.
Some more refs: for implementation, rxinfer.com (toolbox), and arxiv.org/abs/2504.14898 (AIF planning as inference), and lazydynamics.com (company). Also check out noumenal.ai , stanhopeai.com, and verses.ai .
Bert de Vries retweeted
That is 🤯
🚀 What do Bayesian inference and skydiving have in common? Both demand trust under uncertainty. Our CTO @bvdmitri used RxInfer to clean up noisy pose estimates from his 500th skydive — showing how probabilistic inference fills the gaps where standard ML fails #Bayes #Skydiving