Search results for #ProbProg
Tired of waiting forever for MCMC chains to converge? We experimented with using Pathfinder VI to initialize HMC and get early model diagnostics. mlcolab.org/public-events/… #bayesian #probprog #probml
@AndrewC_ML @willarvey @ValentinDeBort1 @tom_rainforth @ArnaudDoucet1 I am down to discuss its connections to graphical models, simulator based inference/ #probprog compilers such as github.com/plai-group/dap… also over a coffee. Feel free to reach out 😀
RT @[email protected] With PosteriorDB.jl v0.3.0, it's easier than ever to load #Stan models from posteriordb for sampling with StanSample.jl. github.com/sethaxen/Poste… #JuliaLang #Bayesian #Statistics #ProbProg #FOSS fosstodon.org/@sethaxen/1092…
Are you interested in composable Bayesian inference, especially in Julia? It would help me to have more active discussions about this for Tilde. Things like •MCMC as iteration •nested models •models as latent+pushforward #bayes #probprog #julialang github.com/cscherrer/Tild…
1997: Started using #Python 2008: ditched cable, got a Roku box and Netflix 2010: Started doing #probprog 2015: started using #julialang
1997: Started using #Python 2008: ditched cable, got a Roku box and Netflix 2010: Started doing #probprog 2015: started using #julialang
Way more interesting for me were the research interviews. These are very subjective, problem-based and topic-focused. This year mine were focused around #causality and #probprog and boy was that fun!
#probprog hot take: For trace-based inference with stochastic control flow, branch traces should be in a sum type. That way, simulating the latents can be strongly typed and type-stable. Partly hot because I haven't fully convinced myself yet 🔥
#probprog book!
International Conference on Probabilistic Programming 2021 (#ProbProg) is online Wed Oct 20 2021 - Fri Oct 22 2021. Here's the schedule: probprog.cc/schedule/ #ProbabilisticProgramming
Looks like big tech interest in probabilistic programming is picking up #probprog
A hierarchical model cannot ‘compress’ training data for efficient inference on future observations. Is there a way to learn from data in probabilistic programs which is both Bayesian in nature and computationally efficient? dtolpin.github.io/posts/h2typ/ #probprog #bayes
@roydanroy That's very surprising, makes me curious about their view on #probprog in particular. Maybe they just see it as a thin wrapper over the numerical algorithms?
@Martin_Trapp "Publications" in academia, "inertia" outside it. Same problem for #julialang , same problem for #probprog
Really impressed so far with @zulip. Like @SlackHQ but - open source (no scrolling into the void) - LaTeX-friendly - better syntax highlighting - better threading model means more bandwidth per channel See you in the #julialang #probprog channel julialang.zulipchat.com
Research data is an essential foundation for scientific work. How can #probprog, #neurosymbolic, and #AI help to understand this rich data? Happy to be appointed to the working group "Handling of Research Data" of the Wissenschaftliche Kommission Niedersachsen. @TUDarmstadt
Style file and guide: it's here! #CallforPapers #CallForSubmissions #ProbProg Web and email update with this info to follow shortly. probprog.cc/2020/probprog-…
"Big tech" investment in probabilistic programming: - @GoogleAI (@TensorFlow probability, Edward) - @FacebookAI (Prophet, HackPPL) - @UberAILabs (Pyro) Others? #probprog
Fun or awful #probprog challenge: Derive a random forest (or similar) as a posterior sample. Prior is a single decision tree on a single bootstrap of the rows. Has anyone done this? Sounds like a good one for @TuringLang
PROBPROG 2021 @probprog
480 Followers 224 Following The International Conference on Probabilistic Programming