Jan Boelts

Researcher at
Maintainer of the Simulation-Based Inference Toolkit @sbi_devs

[Jan Boelts @janfiete]

Model misspecification is one of the trickier challenges in SBI: what happens when your simulator doesn’t capture the observed data?

I wrote an overview of the problem and methods to detect and handle it, now out in the ICLR 2026 blog post track: https://t.co/PuTbZgQKPq

[Jan Boelts @janfiete]

@giswqs This is awesome @giswqs ! How does integrate with custom models, e.g. could I let it use my fine-tuned SAM variant?

[sbi developers @sbi_devs]

From hackathon to release: sbi v0.25 is here! 🎉
What happens when dozens of SBI researchers and practitioners collaborate for a week? New inference methods, new documentation, new embedding networks, a bridge to pyro and a bridge between flow matching and score-based methods 🤯

[Jan Boelts @janfiete]

For everyone working with trial-based i.i.d. data and varying experimental conditions - we have you covered now!
You can train NLE once and then run MCMC with multiple subjects, trials and conditions, etc.
Example: https://t.co/t2AujhvQz2
Reach out on GitHub for questions 🙋‍♂️

[sbi developers @sbi_devs]

🎉 Just in time for the end of the year, we’ve released a new version of sbi!

📦 v0.23.3 comes with exciting features, bug fixes, and docs updates to make SBI smoother and more robust. Check it out! 👇

🔗 Full changelog: https://t.co/bn4qICJMcU

[Jan Boelts @janfiete]

So happy to be part of this project and see it growing! Many thanks to all contributors and users for making it possible 🚀

[sbi developers @sbi_devs]

The sbi package is growing into a community project 🌎 To reflect this and the algorithms, neural nets, and diagnostics that have been added since its initial release, we have written a new software paper. Reach out if you want to get involved: https://t.co/IKRLTONnbj

[Jan Boelts @janfiete]

We have an exciting talk coming up tomorrow at 4pm CEST in the TransferLab seminar! Join us for an introduction to Reduced Order Modeling and how surrogate models can accelerate large-scale engineering workflows. More details below and here: https://t.co/R5a3m8mJPT

[Jan Boelts @janfiete]

Looking for some physics-enhanced machine learning insights this Thursday afternoon? 🧠
Join us for the upcoming TransferLab talk on "generalized quadratic embeddings for nonlinear dynamics".

[Jan Boelts @janfiete]

Absolutely! Huge kudos to @FrancoisRozet for Zuko—it’s incredibly smooth and efficient, also for flow matching! 🚀 Can’t wait to try out Azula next! 🔥

[François Rozet @FrancoisRozet]

So many new features 🤯 Train and sample faster from your (S)NPE by using Zuko as a backend for normalizing flows!

[Jan Boelts @janfiete]

Make sure to check out the fresh look of our docs website: https://t.co/9j2hgEg6Yx🎨
Big thanks to @tomamoral for the incredible help and kudos to https://t.co/pwzFuMfC9y, making docs version switching so seamless.

[Jan Boelts @janfiete]

So excited to see the new SBI release out just in time for today’s talk at #EuroSciPy2024! 🎉 A big shoutout to all the new contributors who made this possible. Thank you all! 🙏 Names are here: https://t.co/ZjWIk55AiF

[sbi developers @sbi_devs]

We just released a new version of `sbi`, and this one has _a ton_ of new features! Many of these features are thanks to more than 30 (mostly new) contributors. We are very excited about the growing community and the new release! 🧵 1/8

[Jan Boelts @janfiete]

Excited to present the SBI toolkit at #EuroSciPy2024 next week! 🎉 I’ll introduce SBI and show how @sbi_devs gives user-friendly access to the latest methods for Bayesian parameter inference on scientific simulators. See you there! #OpenSource #Python

[Jan Boelts @janfiete]

Really happy to see this come to life! 🚀

[sbi developers @sbi_devs]

We’re excited to share that the SBI package is now officially @NumFOCUS affiliated and growing into a community project. A lot of upcoming release originated from a hackathon earlier this year, and there are more events planned soon. Stay tuned for details! #opensource #hackathon

[Jan Boelts @janfiete]

https://t.co/OLCnFtfHSY

[Jan Boelts @janfiete]

🚆 Heading to #icml2024? Need a good read for the journey? Check out the paper summaries I wrote at https://t.co/KjO4KfMNd6! 📚

📝 "All-in-One SBI", outstanding work by Gloeckler et al. (oral) @mackelab
📝 "Position: Leverage LLMs for BBO" by the amazing @RobertTLange et al.

[Jan Boelts @janfiete]

https://t.co/Q0xwLEDGw5

[Machine Learning in Science @mackelab]

Neural networks are highly sensitive to tiny adversarial perturbations. Can we trust them to reliably perform Bayesian inference?
We investigated the brittleness of inference networks & possible solutions. By @gloecklermanuel @deismic_ @jakhmack, accepted at ICML. 🧵1/5

[Machine Learning in Science @mackelab]

We are hiring postdocs!

If you're interested in working at the intersection of machine learning and neuroscience, join us in Tübingen @uni_tue to build biologically realistic and task-performing deep neural networks.

Check out the full job ad here: https://t.co/anNlccgonU

[Jan Boelts @janfiete]

@RobertTLange @mackelab @jakhmack Thank you, Robert!

[Jan Boelts @janfiete]

Finally done 👨‍🎓🥳 Many thanks to the amazing @mackelab and @jakhmack for supporting and guiding me on this journey!

[Machine Learning in Science @mackelab]

New Doctor in the house!

Dr. Jan Boelts @janfiete successfully defended his PhD last week, talking about a fantastic set of works on developing and applying simulation-based inference methods in neuroscience and cognitive science.

Many congratulations for making it to the top!

[Machine Learning in Science @mackelab]

Neural oscillations are abundant! But what is their role? Oscillations could serve as a clock signal for phase-coding information.

In: https://t.co/qxWgSPfPtJ, we studied the dynamical systems underlying this kind of coding. With @matthijs_pals, @jakhmack and Omri Barak [1/5]