iiinova — where your digital self keeps evolving

Hidde Fokkema @hiddefokkema.bsky.social · 15/10/2026

I would say the inverse of open should be open, combined with bijectivity, that would mean that f^-1((0, 1)) = [0, 1), which would not be open. I know that [0, 1) would be open in the subset topology, but I am sure that we can work around that.

110 0

Reposted by Hidde Fokkema

Tim van Erven @timvanerven.nl · 08/10/2026

With AI & Mathematics (AIM) we are writing a draft vision document that we would like to discuss with the Dutch mathematics community. Please join us: AI4Math Workshop: Shaping the Future of Mathematics with AI Date: Friday 4 December 2026 Time: 09:30 - 15:00 Website: ai4math.nl

ai4math.nl

AI4Math Workshop: Shaping the Future of Mathematics with AI

A half-day workshop on developing a shared vision for the future of mathematics in the age of AI. Friday 4 December 2026, 09:30-15:00, CWI Amsterdam.

084 0

Hidde Fokkema @hiddefokkema.bsky.social · 07/05/2026

I looked into rewriting the regression optimization in Cython as well, but it's already quite optimized algorithmically and most operations were vectorized. So the gains weren't worth the added complexity. If anyone has any ideas on this though, I would be very interested to hear your thoughts !

000 0

Hidde Fokkema @hiddefokkema.bsky.social · 07/05/2026

This was also my first time building wheels across multiple platforms and Python versions, so if anything breaks on your setup, I'd genuinely appreciate a heads-up!

100 0

Hidde Fokkema @hiddefokkema.bsky.social · 07/05/2026

Pushed a new update to the regmmd scikit package! ⚡ Performance — A 3–6x speedup on the estimation optimization, thanks to rewriting parts in Cython. 📖 Documentation — Improved the docs. 🐛 Bug fixes — A handful of minor fixes. Check it out → github.com/HiddeFok/reg...

github.com

GitHub - HiddeFok/reg-mmd-scikit: Scikit Implementation of the regMMD estimation and regression procedure

Scikit Implementation of the regMMD estimation and regression procedure - HiddeFok/reg-mmd-scikit

121 0

Hidde Fokkema @hiddefokkema.bsky.social · 27/03/2026

I am not particularly bullish, but I do think that Google will eventually pull ahead, they are the only player that is not entirely dependent on large cash injections through their other revenue streams and I think that eventually the other companies will not be able to cater to their investors

010 0

Hidde Fokkema @hiddefokkema.bsky.social · 20/03/2026

Go check out the package I wrote for the robust regression and estimation procedure Mathie Gerber and @pierrealquier.bsky.social developed based on the Maximum Mean Discrepancy principle!

031 0

Reposted by Hidde Fokkema

Sara Magliacane hiring PhDs in Saarland @smaglia.bsky.social · 13/03/2026

🚨2 PhD positions with me @amlab.bsky.social on learning causally grounded concepts 🚨 Are you interested in improving the #interpretability #robustness and #safety of AI by integrating #causal reasoning? Join us in beautiful Amsterdam 🇳🇱🌷🚲 Deadline: 20 April www.academictransfer.com/en/jobs/3593...

academictransfer.com

2 PhD Positions on Learning Causally Grounded Concepts for Safe AI

Are you interested in improving the interpretability, robustness and safety of AI by integrating causal reasoning? The Causality team in the AMLab group at the University of Amsterdam is looking for 2...

02014 0

Reposted by Hidde Fokkema

Tim van Erven @timvanerven.nl · 09/02/2026

Excited for Damien’s seminar talk this Thursday!🚀 Steering is an exciting area in interpretability—but how strongly should we steer? Damien will present a theory of steering strength: choosing the right magnitude of representation change—not too weak, not too strong tverven.github.io/tiai-seminar/

021 0

Reposted by Hidde Fokkema

Gunnar König @gunnark.bsky.social · 02/12/2025

At #NeurIPS in San Diego this week? Interested in XAI, causality, or performative prediction? Come visit our poster! 💬 Performative Validity of Recourse Explanations 📆 Wednesday, 4.30 pm, Poster Session 2 w/ Hidde Fokkema, Timo Freiesleben, Celestine Mendler-Dünner, Ulrike von Luxburg

0123 0

Reposted by Hidde Fokkema

Tim van Erven @timvanerven.nl · 21/11/2025

When reading a large literature, it is really helpful to have opinionated opinions that help to categorize papers. One of mine for explainable AI is that methods need to address a fundamental limit in how much information can be communicated. Blog post: www.timvanerven.nl/blog/xai-com... (no math)

timvanerven.nl

The Central Challenge in Explainable AI: Channel Capacity

Explainable AI is about communication: we want to tell people how or whya machine learning model is making certain decisions. Why is this sodifficult? In this post I take an information-theoretic pers...

084 0

Hidde Fokkema @hiddefokkema.bsky.social · 03/11/2025

I did some googling and this article has a surprisingly nice and pedagogical discussion on this, with a similar conclusion to your idea. tinyurl.com/52a3whac And I found that I missed the opportunity to make the joke that the posterior of the simpler model is "Sharper", keeping the razor theme.

tinyurl.com

Ockham's Razor and Bayesian Analysis on JSTOR

William H. Jefferys, James O. Berger, Ockham's Razor and Bayesian Analysis, American Scientist, Vol. 80, No. 1 (January-February 1992), pp. 64-72

110 0

Hidde Fokkema @hiddefokkema.bsky.social · 03/11/2025

(2/2) if we see the complicated model and simple model as 2 different hypothesis classes, with 2 seperate priors, then the posterior for the more complicated class will be flatter than the posterior of the simple class, which is what you want I think.

110 0

Hidde Fokkema @hiddefokkema.bsky.social · 03/11/2025

(1/2) Fair point, I think my point was that anything Bayesian is prior related. So with the correct prior you could at least recover Ockham's razor, but not really derive it. But my thinking is a bit different I think, as in my points above the hypothesis classes are the same. In your idea, if ..

120 0

Hidde Fokkema @hiddefokkema.bsky.social · 03/11/2025

(7/n=7) So, in the end, you can get Ockham's razor if your prior is that simple explanations (read explanations with less parameters) are more likely than complicated ones. For binary parameters you could write the prior explicitly. For real valued parameters this becomes impossible (I am guessing)

120 0

Hidde Fokkema @hiddefokkema.bsky.social · 03/11/2025

(6/n) Now if you really want to derive Ockham's razor, in the sense of minimum assumptions, or really the number of parameters, you would need a prior distribution that assigns more probability mass to simple models.

110 0

Hidde Fokkema @hiddefokkema.bsky.social · 03/11/2025

(5/n) Similarly, if β ~ Laplace(0, σ^2), then you get the Lasso objective min ||y - <β, x> ||^2 + λ || β ||_1 where we now have the 1-norm as regularization penalty. This one has the added benefit that irrelevant parameters are set to 0, which resembles the original Ockham's razor principle more

110 0

Hidde Fokkema @hiddefokkema.bsky.social · 03/11/2025

(4/n) writing out the posterior likelihood and performing maximum likelihood on the parameters (Maximum a postiori Bayesian inference). How much you regularise is determined by σ, and there is relation to λ.

120 0

Hidde Fokkema @hiddefokkema.bsky.social · 03/11/2025

(3/n) Let's say we consider as possible models all linear models and complexity measure the euclidean norm of the parameters. (This is ridge regresions). Then, we would retrieve the optimisation problem: min ||y - <β, x>||^2 + λ||β||^2 By assuming that β ~ N(0, σ^2) and ...

120 0

Hidde Fokkema @hiddefokkema.bsky.social · 03/11/2025

(2/n) In particular, this would give you the model with the least amount of assumptions, if you consider 2 models that explain the data equally well, but one has less assumptions and that is the complexity measure you consider.

120 0

Hidde Fokkema @hiddefokkema.bsky.social · 03/11/2025

Sure! Here are some thoughts (1/n) I would see Ockham's razor as the following optimisation problem: min Error(data, model) + Complexity(model) Where you minimise over all models.

120 0

Hidde Fokkema @hiddefokkema.bsky.social · 03/11/2025

If you see Ockham's razor as a regularization mechanism, because you optimize to fit the data and minimizing the parameters, then there are explicit connection. For example ridge regression follows from assuming a gaussian prior on the parameters and Lasso regression follows from a Laplace prior

120 0

Reposted by Hidde Fokkema

Tim van Erven @timvanerven.nl · 13/06/2025

4th AI & Mathematics in NL workshop in Tilburg. Many cool presentations: aimath.nl/index.php/20... And great people:

0143 0

Reposted by Hidde Fokkema

Tim van Erven @timvanerven.nl · 30/05/2025

Deadlines for PhD and Postdoc vacancies coming up: applications open until Monday June 2!

055 0

Reposted by Hidde Fokkema

Tim van Erven @timvanerven.nl · 03/04/2025

Now open: vacancies for a PhD or Postdoc position to develop "Mathematical Foundations for Explainable AI" with me. This is a new research direction that I am very excited about, and which will really start to take off over the next few years. Come join my group: www.timvanerven.nl#open-phd-and...

timvanerven.nl

Tim van Erven

Tim van Erven’s website

0165 0

Reposted by Hidde Fokkema

Rianne de Heide @rdeheide.bsky.social · 14/03/2025

Jelle Goeman (Leiden University Medical Center) and I (University of Twente) have two PhD positions on e-values and multiple testing. The students will be co-supervised by both of us. A strong theoretical mathematical background is required. utwentecareers.nl/en/vacancies...

utwentecareers.nl

Two PhD positions on Flexible and User-adaptive statistical inference - Looking for a job that matters?

You will develop a mathematical framework for multiple testing, enabling flexibility in study design, analysis, and model choice while retaining strong error guarantees. This means that researchers ca...

084 0

Reposted by Hidde Fokkema

Tim van Erven @timvanerven.nl · 01/03/2025

Hooray, I received a Vici grant from the Dutch science foundation! Heads up for current PhD students in learning theory: I will have two postdoc positions available in Amsterdam on "learning theory for interpretable/explainable AI" in the coming years. www.uva.nl/shared-conte...

uva.nl

NWO Vici grants awarded to six UvA/AMC researchers

Six UvA and AMC academics have been awarded Vici grants worth up to €1.5 million by the Dutch Research Council (NWO), to pursue research into topics ranging from black holes to combatting obesity. The...

5325 0

Reposted by Hidde Fokkema

Tim van Erven @timvanerven.nl · 17/02/2025

Reposting to see if we can get some input on what the community is eager to see in the seminar:

063 0

Reposted by Hidde Fokkema

Tim van Erven @timvanerven.nl · 04/02/2025

Great talk by Jeremias Sulam in the interpretable AI seminar today, connecting feature and concept interpretability to hypothesis testing via E-values! Recording available in our YouTube channel: youtu.be/cx7wTtRdhnA Check out the seminar website for upcoming talks: tverven.github.io/tiai-seminar/

youtu.be

Theory of Interpretable AI Seminar: Jeremias Sulam

YouTube video by Theory of Interpretable AI Seminar

052 0

Hidde Fokkema @hiddefokkema.bsky.social · 10/12/2024

Ohw and the timing of the Q2B conference being this week probably also factors in. So they can hype it a bit more there

030 0

Hidde Fokkema @hiddefokkema.bsky.social · 10/12/2024

My guess would be because the nature version of the article was just published?

130 0

Reposted by Hidde Fokkema

Tim van Erven @timvanerven.nl · 03/12/2024

📢Theory of Interpretable AI Seminar📢 On Thursday, Depen Morwani from Harvard will present work on how **margin maximization** can explain observed phenomena in mechanistic interpretability! ⏲️Thursday Dec 5th, 4pm CET / 10am EST 🌐https://tverven.github.io/tiai-seminar/

031 0

Hidde Fokkema @hiddefokkema.bsky.social · 29/11/2024

Aren't these dual numbers? I think Julia has some autodiff packages based in this idea

120 0