Boris Sobolev
@soboleffspaces
𝗦𝗰𝗵𝗼𝗹𝗮𝗿/𝗔𝘂𝘁𝗵𝗼𝗿/𝗧𝗲𝗮𝗰𝗵𝗲𝗿 • causality in plain language •
قد يعجبك
I’m a Pearlian! I follow and advocate for the epistemological approach of Judea Pearl. You can call me Жемчужников.
So are you a realist then? Again, I said framework. The what about it is what would your answer be. If the way I phrased it was imprecise, okay. But if not, your reaction is curious. The fact that regression was created at all would imply it offered some utility.
While bureaucrats and the “concerned citizenry” panic about chatbots, LLMs are busy doing what they can’t — making us more productive. Here’s a figure Overleaf generated from LaTeX code that Gemini 2.5 Pro writes from my verbal (3 sentences) description of this NCM architecture!…
And! Not only did causal theory @yudapearl lay the foundation for the “how” in mediation, fairness, and counterfactual analysis. It also gave us the “how” of Causal AI, opening the era of computable individual causal effects! In the process, it threw the regression…
Is it just me, or is “IRL-FAFO-headcanon” basically Pearl’s “seeing-doing-imagining” in Zoomer lingo? 🤭
If causal knowledge only comes from experiments or experience, then it just tells us how things happen, not how they have to be. But somehow, do-calculus and CTF-calculus manage to be pure reasoning, and still tell us what to expect in the real world.
Frank, a fair challenge in public space: a table in which causality concepts are expressed side by side in the language of DOE and SCM.
You are talking as though causal inference is opposing "experimental designs" etc. -- It isn't. Causal Inference is a language with which one can talk coherently and transparently about all aspect and cause-effect relationships.
The FDA framework for counterfactual estimands - what would have happened to the same patients had they been treated differently - is attempted with statistical machinery like outcome modeling, IPW, and principal strata. The problem is handling intercurrent events, i.e.,…
Implementation of the ICH E9 (R1) addendum in vaccine efficacy studies: the hypothetical and principal stratum strategies tandfonline.com/doi/full/10.10…
Well, ‘Freedom is the recognition of necessity’ is known at least since Hegel. 😊
"When we learn the laws of nature, how certain transformations are not possible, our freedom and potential actually increase, often dramatically so." ~Conjecture Institute Cofounder @astupple
In my #causality course, we spend a lot of time on the notion of a study unit: the real-world entity we intervene on and watch their response. Today I felt poetic: “In the beginning was the Unit. And the Unit was with Treatment, and with Outcome. And each possible Treatment…
Congratulations and thank you for your contribution to the field
My latest contribution to the theory of causal inference, just published in IJE; revisiting propensity scores (Rosenbaum and Rubin, 1983) and disease risk scores (Hansen, 2008) through the lens of causal diagrams (Pearl, 1995): academic.oup.com/ije/article-ab…
Hats off to @soboleffspaces, for bringing causal-ai to medicine. Calling for reforming @NIH's thinking. @NIHDirector_Jay @eliasbareinboim @murat_kocaoglu_ @pierreguyubc @hippysurgeon @JacobJHutton
🚀 Big milestone for our team! We’ve made our first real progress in implementing Causal AI, in the spirit of Pearl & Bareinboim’s SCM and DAG-informed training, to understand the causal effects of care standards in treating patients with hip fracture and to bring individualized…
What’s ahead is already now. Многая лѣта!
Conjecture Institute is honored to announce our second Advisor, computer scientist and philosopher @yudapearl! His books, including Causality and The Book of Why, have influenced fields ranging from epidemiology to economics.
The Guardian does a good job presenting the scientific challenge of linking paracetamol to autism. But the politically-motivated “mistrust” framing hides the sloppy research: • No causal diagram showing dependencies among factors that influence exposure and outcome. • No proven…
An intervention may help some and harm others—an idea conceptualized by @yudapearl at L3 of the causal hierarchy and further developed by @eliasbareinboim into counterfactual calculus.
A Columbia Mailman study provides the first human evidence linking #BreastCancer to a chemical compound once used as an anticancer drug. The research also shows how metabolomics can uncover hidden environmental risks and improve prediction in high-risk women.…
United States الاتجاهات
- 1. Good Wednesday 25K posts
- 2. #LoveYourW2025 168K posts
- 3. #wednesdaymotivation 5,485 posts
- 4. #VxWKOREA 44K posts
- 5. Hump Day 10.2K posts
- 6. Markey N/A
- 7. And the Word 75.8K posts
- 8. #WednesdayWisdom N/A
- 9. #15Oct 3,400 posts
- 10. Moulton N/A
- 11. Happy Hump 6,246 posts
- 12. St. Teresa of Avila 2,009 posts
- 13. Raila Odinga 173K posts
- 14. Tami 5,122 posts
- 15. LEE KNOW FOR HARPERS BAZAAR 6,794 posts
- 16. Young Republicans 94.9K posts
- 17. Kenya 130K posts
- 18. George Floyd 37.7K posts
- 19. Baba 133K posts
- 20. cate 4,760 posts
قد يعجبك
-
CAUSALab
@CAUSALab -
Judea Pearl
@yudapearl -
Elias Bareinboim
@eliasbareinboim -
Peng Ding
@pengding00 -
Sander Greenland
@Lester_Domes -
Análise Real
@analisereal -
Frank Harrell
@f2harrell -
Brady Neal
@CasualBrady -
Mark van der Laan
@mark_vdlaan -
Martin Huber
@CausalHuber -
Joshua Loftus
@joftius -
Jeffrey Wooldridge
@jmwooldridge -
Solomon Kurz
@SolomonKurz -
Sepp Hochreiter
@HochreiterSepp -
Gözde Kuyumcu
@GozdeKuyumcu
Something went wrong.
Something went wrong.