Can conformal prediction help with inference on statistical parameters?
In our new paper, 𝗖𝗼𝗻𝗳𝗼𝗿𝗺𝗮𝗹 𝗖𝗮𝗹𝗶𝗯𝗿𝗮𝘁𝗶𝗼𝗻 𝗼𝗳 𝗦𝘁𝗮𝘁𝗶𝘀𝘁𝗶𝗰𝗮𝗹 𝗖𝗼𝗻𝗳𝗶𝗱𝗲𝗻𝗰𝗲 𝗦𝗲𝘁𝘀, published on @TmlrOrg, we extend conformal ideas beyond prediction and introduce TRUST +
Happy to share our paper "Epistemic Uncertainty in Conformal Scores: A Unified Approach" is now on PMLR! 🎉
Selected for oral presentation at UAI 2025!
Big thanks to @kuben45, Vagner & Thiago for the partnership!
🔗https://t.co/5ZnJVjzSpr
#ML#AI#PMLR#ConformalPrediction
🚨 Thrilled to share one of my favorite papers:
REACT to NHST: Sensible conclusions for meaningful hypotheses — now out in The Quantitative Methods for Psychology! With Luben Cabezas, @FernandoColug, Rodrigo Lassance, @altayals & Rafael Stern.
#Statistics#DataScience
1/n
🚨 Paper accepted for oral presentation at #UAI2025! 🎉
EPICSCORE: A Unified Framework for Incorporating Epistemic Uncertainty in Conformal Scores
Here’s why it matters 🧵
(with amazing co-authors: Vagner S. Santos, Thiago R. Ramos, @rizbicki)
4/
Strong results across tasks 📈
EPICSCORE adapts well to diverse settings—from regression to image classification—while improving uncertainty estimates 🔍✅
Happy to share our work, "Adding Imprecision to Hypotheses: A Bayesian Framework for Testing Practical Significance", with R. Lassance and R. Stern!
We introduce PROTEST, a method for testing practical significance in univariate & high-dimensional data. +
Our paper "Regression Trees for Fast and Adaptive Prediction Intervals," co-authored with @kuben45, @mpotto1 and @rbstern, is now published in Information Sciences! 🎉
We introduce Locart and Loforest to calibrate prediction intervals for regression with coverage guarantees. +
In celebration of its republication by @CRC_MathStats, we are giving away a signed copy of this classic textbook by Casella and Berger. Just like, repost, and follow me by Thursday 15th August to be in with a chance of winning! Enjoy and learn! #Statistics#DataScience#JSM2024
@maria__cuellar @rizbicki@FernandoColug@rflassance@altayals@rbstern Yep, it is named REACT.glm. I've added an example of how to use it in the documentation, but is basically using the vanilla glm with some additional parameters.
O telefone levou 50 anos para ter 50 milhões de usuários. A internet, menos de 10. Com celulares em 2019, Mario Kart levou 7 dias para ter 90M downloads. No que mais a curiosidade vai mudar a sua vida cada vez mais rápido? Descubra: https://t.co/yEv6zsunIT #alwayscurious#ad