Arik Reuter
@arikreuter.bsky.social
University of Cambridge and
Max Planck Institute for Intelligent Systems
I'm interested in amortized inference/PFNs/in-context learning for challenging probabilistic and causal problems.
https://arikreuter.github.io/
Max Planck Institute for Intelligent Systems
I'm interested in amortized inference/PFNs/in-context learning for challenging probabilistic and causal problems.
https://arikreuter.github.io/
Jake @jakemrobertson.bsky.social and I are super excited to share that our paper “Do-PFN: In-Context Learning for Causal Effect Estimation” has been accepted at NeurIPS as a spotlight!
Check out our pre-print on arXiv and stay tuned for the updated version: arxiv.org/abs/2506.06039
[1/7]
Check out our pre-print on arXiv and stay tuned for the updated version: arxiv.org/abs/2506.06039
[1/7]
Do-PFN: In-Context Learning for Causal Effect Estimation
Estimation of causal effects is critical to a range of scientific disciplines. Existing methods for this task either require interventional data, knowledge about the ground truth causal graph, or rely...
arxiv.org
September 25, 2025 at 9:25 AM
Jake @jakemrobertson.bsky.social and I are super excited to share that our paper “Do-PFN: In-Context Learning for Causal Effect Estimation” has been accepted at NeurIPS as a spotlight!
Check out our pre-print on arXiv and stay tuned for the updated version: arxiv.org/abs/2506.06039
[1/7]
Check out our pre-print on arXiv and stay tuned for the updated version: arxiv.org/abs/2506.06039
[1/7]
Reposted by Arik Reuter
New preprint: SBI with foundation models!
Tired of training or tuning your inference network, or waiting for your simulations to finish? Our method NPE-PF can help: It provides training-free simulation-based inference, achieving competitive performance with orders of magnitude fewer simulations! ⚡️
Tired of training or tuning your inference network, or waiting for your simulations to finish? Our method NPE-PF can help: It provides training-free simulation-based inference, achieving competitive performance with orders of magnitude fewer simulations! ⚡️
July 23, 2025 at 2:28 PM
New preprint: SBI with foundation models!
Tired of training or tuning your inference network, or waiting for your simulations to finish? Our method NPE-PF can help: It provides training-free simulation-based inference, achieving competitive performance with orders of magnitude fewer simulations! ⚡️
Tired of training or tuning your inference network, or waiting for your simulations to finish? Our method NPE-PF can help: It provides training-free simulation-based inference, achieving competitive performance with orders of magnitude fewer simulations! ⚡️
Reposted by Arik Reuter
Can transformers learn full Bayesian inference in context? 🤔
👉 Come find out and visit our poster E-1205 in the East Hall this morning at #ICML2025 presented by @arikreuter.bsky.social
icml.cc/virtual/2025...
👉 Come find out and visit our poster E-1205 in the East Hall this morning at #ICML2025 presented by @arikreuter.bsky.social
icml.cc/virtual/2025...
ICML Poster Can Transformers Learn Full Bayesian Inference in Context?ICML 2025
icml.cc
July 16, 2025 at 2:47 PM
Can transformers learn full Bayesian inference in context? 🤔
👉 Come find out and visit our poster E-1205 in the East Hall this morning at #ICML2025 presented by @arikreuter.bsky.social
icml.cc/virtual/2025...
👉 Come find out and visit our poster E-1205 in the East Hall this morning at #ICML2025 presented by @arikreuter.bsky.social
icml.cc/virtual/2025...
Reposted by Arik Reuter
Compute is increasing much faster than data. How can we improve classical supervised learning long term (the underlying tech of most of GenAI)?
Our ICML position paper's answer: simply train on a bunch of artificial data (noise) and only do inference on real-world data! 1/n
Our ICML position paper's answer: simply train on a bunch of artificial data (noise) and only do inference on real-world data! 1/n
July 8, 2025 at 8:03 PM
Compute is increasing much faster than data. How can we improve classical supervised learning long term (the underlying tech of most of GenAI)?
Our ICML position paper's answer: simply train on a bunch of artificial data (noise) and only do inference on real-world data! 1/n
Our ICML position paper's answer: simply train on a bunch of artificial data (noise) and only do inference on real-world data! 1/n
We present a new approach to causal inference. Pre-trained on synthetic data, Do-PFN opens the door to a new domain: PFNs for causal inference—we are excited to announce our new paper “Do-PFN: In-Context Learning for Causal Effect Estimation” on Arxiv! 🔨🔍
A thread: 🧵[1/8]
A thread: 🧵[1/8]
June 10, 2025 at 9:33 AM
We present a new approach to causal inference. Pre-trained on synthetic data, Do-PFN opens the door to a new domain: PFNs for causal inference—we are excited to announce our new paper “Do-PFN: In-Context Learning for Causal Effect Estimation” on Arxiv! 🔨🔍
A thread: 🧵[1/8]
A thread: 🧵[1/8]
Reposted by Arik Reuter
It seems that we have 3 accepted papers at ICML 2025 🔥
May 1, 2025 at 12:32 PM
It seems that we have 3 accepted papers at ICML 2025 🔥
Reposted by Arik Reuter
Arriving in Singapore this afternoon 🛬 I'll attend #ICLR2025, #AABI2025, and #AISTATS2025 together with many of my students and collaborators to present our 2 orals, 5 posters, and 14 workshop contributions 🚀
Feel free to drop by!
Feel free to drop by!
April 23, 2025 at 3:15 AM
Arriving in Singapore this afternoon 🛬 I'll attend #ICLR2025, #AABI2025, and #AISTATS2025 together with many of my students and collaborators to present our 2 orals, 5 posters, and 14 workshop contributions 🚀
Feel free to drop by!
Feel free to drop by!
Reposted by Arik Reuter
BREAKING: People are being suspended on X in Turkey for posting videos of these protests against Erdoğan’s corrupt and repressive regime.
Keep sharing everywhere.
Keep sharing everywhere.
March 23, 2025 at 12:20 PM
BREAKING: People are being suspended on X in Turkey for posting videos of these protests against Erdoğan’s corrupt and repressive regime.
Keep sharing everywhere.
Keep sharing everywhere.