Piersilvio De Bartolomeis

PhD student @ETH || Machine Learning and Causal Inference

[Riccardo De Santi @r_de_santi]

Generative models can’t discover what they can’t reach.

We’re excited to introduce ActFlow: a continued pre-training scheme that actively expands the valid design space reachable by flow and diffusion models. We call this generable set expansion — a new learning principle for out-of-distribution generative modeling, and a step toward evolvable search spaces for scientific discovery. (1/5)

[Ilker Demirel @ilkerdemirel_]

I'm at ICML presenting

📜Uncovering Bias Mechanisms in Observational Studies!

Joint work with @zeshanmh, @pdebartols, and @david_sontag.

Come find our poster at:
📌HALL A #4111
🕐 Wed, Jul 8, 5:00 PM – 6:45 PM KST.

Feel free to send a DM to meetup!

1/5

[Piersilvio De Bartolomeis @pdebartols]

@giffmana Too bad the view is buried under clouds most of the time

[Erfan Loghmani @LoghmaniErfan]

Amazing panel today @CauScien! Enjoyed @nathankallus insights on the future of causality + data science in tech: Causality and experimental insights on LLM use across industries, alignment with user feedback, combing data sources to debias and reduce variance, and more...

[CauScien @CauScien]

🚀 Tomorrow is the day!

CauScien @ #NeurIPS2025 starts 8:15 AM -- a full day of causal reasoning, science, and mind-expanding talks.

Come early. Bring curiosity. Leave inspired ✨

[Piersilvio De Bartolomeis @pdebartols]

Check out our blog post for more: https://t.co/HmeGtTYYfN

I’d be very happy to chat at NeurIPS if you’re interested in causal inference, AI for science, and (tabular) foundation models to make randomized experiments more efficient.

[Piersilvio De Bartolomeis @pdebartols]

The key point: you can get tighter confidence intervals while preserving valid statistical inference, even if the model predictions are arbitrarily biased.

We demonstrate that H-AIPW works very well in practice on several social-science experiments (see table below).

[Piersilvio De Bartolomeis @pdebartols]

We introduce Hybrid Augmented Inverse Probability Weighting (H-AIPW), an estimator that safely integrates predictions from foundation models into randomized experiments.

[Piersilvio De Bartolomeis @pdebartols]

Can foundation models make randomized experiments more precise—without breaking inference guarantees?

Excited to present our NeurIPS 2025 paper “Efficient Randomized Experiments Using Foundation Models”

⏳ Wednesday, 11:00 AM – 2:00 PM PT
📍 Exhibit Hall C,D,E #2503

[CauScien @CauScien]

Meet our speaker! 🎙️ @nathankallus (Cornell & Netflix) will speak at #CauScien #NeurIPS2025 on

"Learning Surrogate Indices from Historical A/Bs: Adversarial ML for Debiased Inference on Functionals of Ill-Posed Inverses"

📅Sat Dec 6 | 🕑2:00-2:30pm |📍Upper Level Room 8

[CauScien @CauScien]

#NeurIPS2025 is approaching!!! 🔥

The list of accepted papers is out (link below). See anything that sparks your curiosity? 👀

Come chat with the authors — and a bunch of us thinking about causality × science — at the #CauScien Workshop on Dec 6, Upper Level Room 8.

[NEJM AI @NEJM_AI]

Editorial by Issa J. Dahabreh, MD, ScD, Robert W. Yeh, MD, MSc, MBA (@rwyeh), and Piersilvio De Bartolomeis, MSc (@pdebartols): Trial Emulation, Simulation, and Augmentation Using Electronic Health Records and Generative AI https://t.co/yRfAb0DmTq

#AIinMedicine

[Piersilvio De Bartolomeis @pdebartols]

Submissions are now open! 🚀

If you're exploring causality in science we’d love to see your work.

[Dingling Yao @Dingling_Yao]

🚀 Got fresh ideas in causal discovery, inference, or reasoning for scientific problems? Share them at the @CauScien workshop @ #NeurIPS2025! 📝✨

Submit your papers by 22 Aug 2025 →  https://t.co/uPh5nKg3nZ

[Piersilvio De Bartolomeis @pdebartols]

Excited to announce our workshop on Causality in Science at #NeurIPS2025!

See you in San Diego 🌴🇺🇸

[CauScien @CauScien]

🚨 We’re thrilled to announce our NeurIPS 2025 workshop:
CausCien: Uncovering Causality in Science 🔍✨
We’re uniting ML + science communities to explore how causal learning advances:
🌿 Ecology 🧬 Biology 📊 Social Science & more!

📝 Submit by Aug 22
👉 https://t.co/2elChnmwZk

[Fanny Yang @FannyYangETH]

Register now (first-come first-served) for the "Math of Trustworthy ML workshop" at #LagoMaggiore, Switzerland, Oct 12-16 this year, with a great speaker lineup and the opportunity to present your work as a poster session or contributed talk. Details @ https://t.co/zpTgLgPxQo

[Piersilvio De Bartolomeis @pdebartols]

RAMEN: https://t.co/Wv2pSwGJ00

[Piersilvio De Bartolomeis @pdebartols]

Landed in Singapore for #ICLR—excited to see old & new friends! I’ll be presenting:

📌 RAMEN @ Main Conference on Saturday 10 am (@JavierAbadM @yixinwang_ @FannyYangETH)
📌 Causal Lifting @ XAI4Science Workshop on Sunday (@riccardocadeii @ilkerdemirel_ @FrancescoLocat8 )

[Piersilvio De Bartolomeis @pdebartols]

@JavierAbadM @yixinwang_ @FannyYangETH @riccardocadeii @ilkerdemirel_ @FrancescoLocat8 Causal Lifting: https://t.co/nXb2ADRb8K

[Piersilvio De Bartolomeis @pdebartols]

@KattaSrikar Congrats!

[Riccardo Cadei @riccardocadeii]

Foundational models’ predictions (🦙♊️🦖) can propagate biases in causal downstream tasks, posing a significant risk in AI-supported scientific discovery (👩‍🔬🔎).

Our solution: Causal Lifting of Neural Representation 🪜