iiinova — where your digital self keeps evolving

Paul Bürkner @paulbuerkner.com · 01/06/2025

the logging is done in rstan. so a fix if needed will have to be there I assume.

050

Paul Bürkner @paulbuerkner.com · 23/05/2025

there is not unfortunately. I didn't have time to look into it anymore.

120

Reposted by Paul Bürkner

Marvin Schmitt @marvin-schmitt.com · 27/03/2025

I defended my PhD last week ✨ Huge thanks to: • My supervisors @paulbuerkner.com @stefanradev.bsky.social @avehtari.bsky.social 👥 • The committee @ststaab.bsky.social @mniepert.bsky.social 📝 • The institutions @ellis.eu @unistuttgart.bsky.social @aalto.fi 🏫 • My wonderful collaborators 🧡 #PhDone 🎓

Image of a graduating PhD student in the trending Studio Ghibli style.

181202

Paul Bürkner @paulbuerkner.com · 21/03/2025

can you post a reprex on github?

100

Reposted by Paul Bürkner

Hadley Wickham @hadley.nz · 27/02/2025

What advice do folks have for organising projects that will be deployed to production? How do you organise your directories? What do you do if you're deploying multiple "things" (e.g. an app and an api) from the same project?

2610029

Reposted by Paul Bürkner

Marvin Schmitt @marvin-schmitt.com · 11/02/2025

Amortized inference for finite mixture models ✨ The amortized approximator from BayesFlow closely matches the results of expensive-but-trustworthy HMC with Stan. Check out the preprint and code by @kucharssim.bsky.social and @paulbuerkner.com👇

0171

Reposted by Paul Bürkner

BayesFlow @bayesflow.org · 11/02/2025

Finite mixture models are useful when data comes from multiple latent processes. BayesFlow allows: • Approximating the joint posterior of model parameters and mixture indicators • Inferences for independent and dependent mixtures • Amortization for fast and accurate estimation 📄 Preprint 💻 Code

0296

Reposted by Paul Bürkner

Aki Vehtari @avehtari.bsky.social · 06/02/2025

If you know simulation based calibration checking (SBC), you will enjoy our new paper "Posterior SBC: Simulation-Based Calibration Checking Conditional on Data" with Teemu Säilynoja, @marvinschmitt.com and @paulbuerkner.com arxiv.org/abs/2502.03279 1/7

Title: Posterior SBC: Simulation-Based Calibration Checking Conditional on Data

Authors: Teemu Säilynoja, Marvin Schmitt, Paul Bürkner, Aki Vehtari

Abstract: Simulation-based calibration checking (SBC) refers to the validation of an inference algorithm and model implementation through repeated inference on data simulated from a generative model. In the original and commonly used approach, the generative model uses parameters drawn from the prior, and thus the approach is testing whether the inference works for simulated data generated with parameter values plausible under that prior. This approach is natural and desirable when we want to test whether the inference works for a wide range of datasets we might observe. However, after observing data, we are interested in answering whether the inference works conditional on that particular data. In this paper, we propose posterior SBC and demonstrate how it can be used to validate the inference conditionally on observed data. We illustrate the utility of posterior SBC in three case studies: (1) A simple multilevel model; (2) a model that is governed by differential equations; and (3) a joint integrative neuroscience model which is approximated via amortized Bayesian inference with neural networks.

45015

Reposted by Paul Bürkner

BayesFlow @bayesflow.org · 03/02/2025

A study with 5M+ data points explores the link between cognitive parameters and socioeconomic outcomes: The stability of processing speed was the strongest predictor. BayesFlow facilitated efficient inference for complex decision-making models, scaling Bayesian workflows to big data. 🔗Paper

0226

Paul Bürkner @paulbuerkner.com · 29/01/2025

cool idea! I will think about how to achieve something like this. can you open an issue on GitHub so I don't forget aboht it?

110

Reposted by Paul Bürkner

BayesFlow @bayesflow.org · 28/01/2025

Join us this Thursday for a talk on efficient mixture and multilevel models with neural networks by @paulbuerkner.com at the new @approxbayesseminar.bsky.social!

0124

Reposted by Paul Bürkner

Approximate Bayes Seminar @approxbayesseminar.bsky.social · 14/01/2025

Paul Bürkner (TU Dortmund University), will give our next talk. This will be about "Amortized Mixture and Multilevel Models", and is scheduled on Thursday the 30th January at 11am. To receive the link to join, sign up at listserv.csv.warwick...

0184

Reposted by Paul Bürkner

Marvin Schmitt @marvin-schmitt.com · 14/01/2025

Paul Bürkner (@paulbuerkner.com) will talk about amortized Bayesian multilevel models in the next Approximate Bayes Seminar on January 30 ⭐️ Sign up to the seminar’s mailing list below to get the meeting link 👇

1214

Reposted by Paul Bürkner

Christian Odendahl @codendahl.bsky.social · 10/01/2025

More than 60 German universities and research outfits are announcing that they will end their activities on twitter. Including my alma mater, the University of Münster. HT @thereallorenzmeyer.bsky.social nachrichten.idw-online.de/2025/01/10/h...

nachrichten.idw-online.de

Hochschulen und Forschungsinstitutionen verlassen Plattform X - Gemeinsam für Vielfalt, Freiheit und Wissenschaft

19541133

Reposted by Paul Bürkner

Riccardo Fusaroli @fusaroli.eurosky.social · 10/01/2025

what are your best tips to fit shifted lognormal models (in #brms / Stan)? I'm using: - checking the long tails (few long RTs make the tail estimation unwieldy) - low initial values for ndt - careful prior checks - pathfinder estimation of initial values still with increasing data, chains get stuck

275

Paul Bürkner @paulbuerkner.com · 09/01/2025

I think this should be documented in the brms_families vignette. perhaps you can double check if the information you are looking for is indeed there.

000

Paul Bürkner @paulbuerkner.com · 03/01/2025

happy to work with you on that if we find the time :-)

020

Paul Bürkner @paulbuerkner.com · 02/01/2025

indeed, I saw it at StanCon but I am not sure anymore how production ready the method was.

020

Paul Bürkner @paulbuerkner.com · 02/01/2025

something like this, yes. but ensuring the positive definiteness of arbitrary constraint correlation matrices is not trivial. so there may need to be some restrictions of what correlation patterns are allowed.

130

Paul Bürkner @paulbuerkner.com · 23/12/2024

I already thought about this. a complete SEM syntax in brms would support selective error correlations by generalizing set_rescor()

260

Reposted by Paul Bürkner

Jarrett Byrnes @jebyrnes.bsky.social · 21/12/2024

OK, here is a very rough draft of a tutorial for #Bayesian #SEM using #brms for #rstats. It needs work, polish, has a lot of questions in it, and I need to add a references section. But, I think a lot of folk will find this useful, so.... jebyrnes.github.io/bayesian_sem... (use issues for comments!)

jebyrnes.github.io

Full Luxury Bayesian Structural Equation Modeling with brms

922560

Reposted by Paul Bürkner

Bruno Hebling Vieira @bhvieira.github.io · 17/12/2024

Ternary plots to represent data in a simplex, yay or nay? #stats #statistics #neuroscience

A ternary plot visualizing data points classified into three groups: CN (Cognitively Normal), MCI (Mild Cognitive Impairment), and AD (Alzheimer's Disease dementia). The triangular axes represent predicted probability corresponding to each diagnosis category, with the corners labeled CN (top), MCI (bottom left), and AD (bottom right). Each point is colored according to its real diagnosis group (blue for CN, green for MCI and red for AD). Trajectories of predicted probabilities pertaining to the same subject are denoted with arrows connecting the points. Shaded regions in blue, green, and orange further highlight distinct areas of the plot associated with CN, MCI, and AD, respectively. A legend on the right identifies the diagnosis categories by color.

3142

Reposted by Paul Bürkner

Kevin M. Kruse @kevinmkruse.bsky.social · 12/12/2024

Writing is thinking. It’s not a part of the process that can be skipped; it’s the entire point.

15667821771

Reposted by Paul Bürkner

BayesFlow @bayesflow.org · 10/12/2024

1️⃣ An agent-based model simulates a dynamic population of professional speed climbers. 2️⃣ BayesFlow handles amortized parameter estimation in the SBI setting. 📣 Shoutout to @masonyoungblood.bsky.social & @sampassmore.bsky.social 📄 Preprint: osf.io/preprints/ps... 💻 Code: github.com/masonyoungbl...

0426

Reposted by Paul Bürkner

BayesFlow @bayesflow.org · 06/12/2024

Neural superstatistics are a framework for probabilistic models with time-varying parameters: ⋅ Joint estimation of stationary and time-varying parameters ⋅ Amortized parameter inference and model comparison ⋅ Multi-horizon predictions and leave-future-out CV 📄 Paper 1 📄 Paper 2 💻 BayesFlow Code

0203

Paul Bürkner @paulbuerkner.com · 07/12/2024

I think the link you cited points to the wrong paper.

110

Paul Bürkner @paulbuerkner.com · 05/12/2024

yeah indeed it seems we don't have it yet. but perhaps may be worthwhile to implement? I will ask on the Stan forums.

020

Paul Bürkner @paulbuerkner.com · 05/12/2024

I don't know. didn't check yet

110

Paul Bürkner @paulbuerkner.com · 04/12/2024

I am always looking for count data distributions that can handle both under and overdispersion without being a computational nightmare. PRs are welcome :)

131

Reposted by Paul Bürkner

Robert Aboukhalil @robert.bio · 04/12/2024

“We don’t value software, data, and methods in the same way we value papers, even though those resources empower millions of scientists” 💯 www.statnews.com/sponsor/2024...

statnews.com

New report highlights the scientific impact of open source software

Two of the scientists who won this year’s Nobel Prize for cracking the code of proteins’ intricate structures relied, in part, on a series of computing

316172

Reposted by Paul Bürkner

Veerle Eeftink - van Leemput @veerle.hypebright.nl · 03/12/2024

The public beta version of Positron was released almost 6 months ago, and the team certainly hasn’t been idle! So what happened over the last half year? Is it worth switching? 👀 I definitely like where it's heading! Personal highlights: data explorer, command palette, help on hover + extensions 👇🏻📚

8345

Reposted by Paul Bürkner

Sean Pinkney @spinkney.bsky.social · 02/12/2024

I'm going to have time to do 1-2 contributions for the summer 25 release. Here's my list to choose from, what is most interesting to you? - adding lower/upper bounds to ordered vectors (removing positive ordered since it's achieved by lb=0)

2143

Paul Bürkner @paulbuerkner.com · 28/11/2024

would you mind adding me there? :)

000

Paul Bürkner @paulbuerkner.com · 27/11/2024

I think it should be possible but I haven't checked the details of the method.

110

Paul Bürkner @paulbuerkner.com · 27/11/2024

I don't know. does the partR2 package support brms models?

120

Paul Bürkner @paulbuerkner.com · 26/11/2024

I am sorry the abstract isnt so clear in that regard. we are currently updating our website of the related software and planning a tutorial paper, which should make this more clear.

110

Paul Bürkner @paulbuerkner.com · 26/11/2024

thank you! we also cite your paper im our related work section. I previously discussed with Petrus and Arto about your and our approaches so they should be aware that the paper is coming eventually.

010

Paul Bürkner @paulbuerkner.com · 26/11/2024

prior eliciation is essentially transforming expert knowledge, provided in whatever form, into a mathematical presentation (our prior distribution) usable for inference. so it's not really empirical bayes since the latter tends to use the actual data not expert knowledge.

120

Paul Bürkner @paulbuerkner.com · 26/11/2024

Did you mean prior elicitation instead of explicitation?

100

Paul Bürkner @paulbuerkner.com · 26/11/2024

not at this stage. but we are planning follow up studies with real experts. the purpose of this paper was to develop the method. but we did try to keep the real experts already in mind.

010

Paul Bürkner @paulbuerkner.com · 26/11/2024

Prior specification is one of the hardest tasks in Bayesian modeling. In our new paper, we (Florence Bockting, @stefanradev.bsky.social and me) develop a method for expert prior elicitation using generative neural networks and simulation-based learning. arxiv.org/abs/2411.15826

arxiv.org

Expert-elicitation method for non-parametric joint priors using normalizing flows

We propose an expert-elicitation method for learning non-parametric joint prior distributions using normalizing flows. Normalizing flows are a class of generative models that enable exact, single-step...

514731

Reposted by Paul Bürkner

Hadley Wickham @hadley.nz · 25/11/2024

I'm writing up a brief history of the tidyverse? What do you want to know about it? #rstats

83521111

Reposted by Paul Bürkner

BayesFlow @bayesflow.org · 25/11/2024

Any single analysis hides an iceberg of uncertainty. Sensitivity-aware amortized inference explores the iceberg: ⋅ Test alternative priors, likelihoods, and data perturbations ⋅ Deep ensembles flag misspecification issues ⋅ No model refits required during inference 🔗 openreview.net/forum?id=Kxt...

1245

Paul Bürkner @paulbuerkner.com · 24/11/2024

I am sorry I don't quite understand the question. can you elaborate further?

100

Reposted by Paul Bürkner

Christoph Molnar @christophmolnar.bsky.social · 24/11/2024

I feel like not enough people know about Quarto for creating documents. How it works: Write in markdown and use Quarto to convert it to html, pdf, epub, ... I produce my books with Quarto (web + ebook + print version). But you can also use it for websites, reports, dashboards, ... quarto.org

quarto.org

Quarto

An open source technical publishing system for creating beautiful articles, websites, blogs, books, slides, and more. Supports Python, R, Julia, and JavaScript.

1618026

Reposted by Paul Bürkner

Marvin Schmitt @marvin-schmitt.com · 24/11/2024

Optimist: The cup is half full Pessimist: The cup is half empty Frequentist: *takes a deep breath* The probability that the cup is half full given the observed volume of water (or more extreme volumes) is larger than 5% so I cannot reject the null hypothesis that the cup is half full.

45010

Reposted by Paul Bürkner

Rachael Shaw @wild-minds.bsky.social · 24/11/2024

Optimist: the cup is half full Pessimist: the cup is half empty Comparative cognition researcher: I wonder if this animal will drop some stones into this cup

0174

Reposted by Paul Bürkner

It's The Weekend 😌 @craigweekend.bsky.social · 23/11/2024

Ladies and gentlemen... the weekend. (also: you are important and are not alone 🧡)

272081588

Reposted by Paul Bürkner

BayesFlow @bayesflow.org · 22/11/2024

BayesFlow is a library for amortized Bayesian inference with neural networks. ⋅ Multi-backend via Keras 3: Use PyTorch, TensorFlow, or JAX. ⋅ Modern nets: Flow matching, diffusion, consistency models, normalizing flows, transformers ⋅ Built-in diagnostics and plotting 🔗 github.com/bayesflow-or...

110822