🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

💡New preprint: Causal Fairness under Unobserved Confounding: A Neural Sensitivity Framework 👉Methods for #causal #fairness assume all variables to be known (≠practice) 🔨We analyze the sensitivity to unobserved confounding arxiv.org/abs/2311.18460 w/ Maresa Dennis Frauen @ICLR 2025

Are you at NeurIPS Conference? Check out the latest papers on #CausalML 🔥 (Spoiler: two of ours are featured 🙃) towardsdatascience.com/causal-python-…

🎉 Our group has 4 papers accepted at ICLR 2026 "A Neural Framework for Generalized Causal Sensitivity Analysis" 📰arxiv.org/abs/2311.16026 👪Dennis Frauen @ICLR 2025 Fergus Imrie Alicia Curth Valentyn Melnychuk 🇺🇦 Mihaela van der Schaar /🧵

I've worked tirelessly on cracking this problem with generations of students. It’s an extremely challenging area of ML and seeing it come closer to clinical use, where it will empower clinicians & patients alike, is very satisfying: nature.com/articles/s4159… #AI #PatientEmpowerment

🚨Our new paper Nature Medicine about #Causal #ML for predicting treatment outcomes Web version: nature.com/articles/s4159… Full access (free): rdcu.be/dFb9K A summary of why it matters 👇

Causal #machinelearning could predict treatment outcomes for subgroups and even individual patients; our latest Perspective outlines the potential benefits and limitations, offering guidance for clinical use. Stefan Feuerriegel Universität München The Alan Turing Institute nature.com/articles/s4159…

I will be at #ICLR2024 in Vienna next week. Please reach out if you would like to meet up!

1/5 Want to know more about representation learning for treatment effect estimation? Come check out our spotlight poster @ ICLR 2024 ICLR 2026 "Bounds on Representation-Induced Confounding Bias for Treatment Effect Estimation" w/ Dennis Frauen @ICLR 2025 Stefan Feuerriegel AI@LMU

🚀New preprint on #causalml for #timeseries: G-Transformer for Conditional Average Potential Outcome Estimation over Time arxiv.org/abs/2405.21012 with Konstantin Dennis Frauen @ICLR 2025 Valentyn Melnychuk 🇺🇦

📣New paper on #causalML ICML Conference: Meta-Learners for Partially-Identified Treatment Effects Across Multiple Environments 👉We estimate treatment effects from multiple🏥 💡Here: overlap & unconfoundedness cancel out arxiv.org/abs/2406.02464 w/ Jonas Dennis Frauen @ICLR 2025 Mihaela van der Schaar

⚡️New preprint: 𝗖𝗼𝗻𝗳𝗼𝗿𝗺𝗮𝗹 𝗣𝗿𝗲𝗱𝗶𝗰𝘁𝗶𝗼𝗻 𝗳𝗼𝗿 𝗖𝗮𝘂𝘀𝗮𝗹 𝗘𝗳𝗳𝗲𝗰𝘁𝘀 𝗼𝗳 𝗖𝗼𝗻𝘁𝗶𝗻𝘂𝗼𝘂𝘀 𝗧𝗿𝗲𝗮𝘁𝗺𝗲𝗻𝘁𝘀 arxiv.org/abs/2407.03094 w/ Maresa Dennis Frauen @ICLR 2025 Jonas Konstantin Valentyn Melnychuk 🇺🇦

Meta-learners for CATE over time! Check out our new preprint

Poster Session 4 (13:30CEST) sees two van der Schaar lab contributions! Drop by to first discuss “Meta-Learners for Partially-Identified Treatment Effects Across Multiple Environments”: a deep dive into the importance of estimating conditional average treatment effects (CATE) 1/3

I am attending NeurIPS 2024! Please reach out if you would like to connect/ meet up

Ever trusted a metric that works great on average, only for it to fail in your specific use case? In our #NAACL2025 paper (w/ Fernando Diaz), we show why global evaluations are not enough and why context matters more than you think. 📄 aclanthology.org/2025.findings-… #NLP #Evaluation (🧵1/9)