#probabilisticprogramming — Public Fediverse posts

// nerd-mode

A small step for the humankind, a huge celebration for my inner nerd.  The most beautiful formulation of Fisher's 1922 dilution assay model in the history of humankind has landed 😊.

A lot has been happening in Probula lately, but chiefly I worked on syntax to allow expressing more complex models, and doing this concisely and elegantly.  The picture shows Fisher's (1922) dilution assay model.  The first non-trivially-structured model that I implemented in Probula; also one whose structure is parametric (depends on the number of dilution stages).

The screenshot shows the entire model, including the inference call. Each line corresponds directly to the piece of math it implements, so the model is almost identical to what McCullagh and Nelder (1989) describe. I think this is amazing. And of course samples very well :) The posterior sd matches Fisher's asymptotic formula to a fraction of a percent.

I will be writing more about this online soon, and wrapping up a Probula release, before my students hit the topic this term.

#scala #scala3 #ScalaLang #probula #ppl #ProbabilisticProgramming #Fisher1922 #statistics #glm #keep

Links: To find this model on codeberg search for FisherIntegrated here https://codeberg.org/wasowski/probula/src/branch/main/doc/example/Fisher1922.scala

//nerd-out

Made an introductory 📕(draft) about using Python for Bayesian Inference and unifying narrative, math, and code. People seem to find it helpful. Check it out. Feedback encouraged.

https://persuasivepython.com

#DataScience #Python #bayes #Stats #probabilisticprogramming

I’m in #Vancouver for the next month. If someone wants meet for chats on #probabilistic #MachineLearning, #ProbabilisticCircuits, or #ProbabilisticProgramming feel free to DM me!

Automatic differentiation in Prolog. ~ Tom Schrijvers, Birthe van den Berg, Fabrizio Riguzzi. https://arxiv.org/abs/2305.07878 #Prolog #LogicProgramming #AutomaticDifferentiantion #ProbabilisticProgramming

Last day of #bayescomp2023, I very much enjoyed yesterday’s panel on #ProbabilisticProgramming. Looking forward to today’s schedule.

This one became a nice example of implementing a custom CUDA kernel through various stages of optimization #CUDA #GPU #ProbabilisticProgramming https://indii.org/blog/sum-of-discrete/

I am looking for
a) examples of tools that let you build statistical models more complex then just variations of a single model class (like most stat packages - brms, laavan, ...) but less complex than fully fledged probabilistic programming languages
b) Probabilistic programming languages that neatly support composing non-trivial submodels together

Does anyone have recs?
In both cases I am coming up almost empty handed...

#stan #ppl #ProbabilisticProgramming #brms

Working on a new CUDA kernel for enumerating sums of discrete random variables #GPU #CUDA #Cpp :cpp_language: #ProbabilisticProgramming https://indii.org/blog/fast-discrete-enumeration/

I had an interesting conversation about #BNP models in #PPLs today, discussing along other things whether truncation is enough (obvious no).

What are your thoughts about BNP priors in #ProbabilisticProgramming? And how to improve automated interference for them?

#bayesian #MachineLearning

CUDA-enabled packages for Birch are now available! (Previously it was necessary to compile from source for CUDA support.) #GPU #CUDA #MachineLearning #ProbabilisticProgramming #MonteCarlo https://birch.sh/getting-started/

On this week's #blog I unravel the great pile-up of acronyms that is 'McNetKAT' https://updatedscholar.blogspot.com/2023/02/discussing-scalable-verification-of.html #atp #verification #networks #ProbabilisticProgramming

We are thrilled to welcome Hugo Paquet as an #MSCA fellow, on a two year Paris Region Fellowship Program.
He is an expert in semantics for programming languages and will work within the LoVe team, on exploiting linearity in probabilistic programming. #ProbabilisticProgramming #Logic
🔗 https://www.cs.ox.ac.uk/people/hugo.paquet/main.html

In 3 weeks, our #Conservation paper has 687 full-text reads! (Incl. ones on researchgate).

I think one of the strengths is using #ProbabilisticProgramming for small populations, letting us do #GWAS and #PopGen better and providing useful estimates back to the conservation practitioners. Lots of other strengths too (high-quality SNP calling) and correcting the record of analyses based off of incomplete genome annotation!

https://www.biorxiv.org/content/10.1101/2022.10.22.513130v1

#introduction

Hi everyone 👋

I'm an Assistant Professor at Utrecht University's dept. of Methodology & Statistics, working in data science for the social sciences.

I do not post a lot, but when I do it's likely about #statistics #datascience #probabilisticprogramming #syntheticdata #openscience #rstats or #teaching

I'm also the (proud) team lead of the ODISSEI Social Data Science team: https://odissei-soda.nl. So I'll probably post about all the cool things our team does too!

The #Stan #ProbabilisticProgramming project is looking for new members of the governance body. If you care about the future of #bayesian computation, consider applying - all types of expertise (user, educator, developer, ...) are highly welcome: https://discourse.mc-stan.org/t/nominations-for-2023-stan-governing-body/29258