Kyle Cranmer
@kylecranmer.bsky.social
Director Data Science Institute @UWMadison, Professor of Physics,
EiC @MLSTjournal. Physics, stats/ML/AI, open science.
EiC @MLSTjournal. Physics, stats/ML/AI, open science.
I am incredibly honored to have received the inaugural AI in Science Research Excellence Prize from the Margot and Tom Pritzker Foundation
dsi.wisc.edu/2025/11/10/d...
dsi.wisc.edu/2025/11/10/d...
November 10, 2025 at 7:22 PM
I am incredibly honored to have received the inaugural AI in Science Research Excellence Prize from the Margot and Tom Pritzker Foundation
dsi.wisc.edu/2025/11/10/d...
dsi.wisc.edu/2025/11/10/d...
🎉@AnimaAnandkumar kicks off the Conference in AI+Science
aiscienceconference.caltech.edu
aiscienceconference.caltech.edu
November 10, 2025 at 4:53 PM
🎉@AnimaAnandkumar kicks off the Conference in AI+Science
aiscienceconference.caltech.edu
aiscienceconference.caltech.edu
Good morning Pasadena
November 10, 2025 at 2:40 PM
Good morning Pasadena
Exciting! Dark Energy Survey collaboration releases first paper demonstrating power of simulation-based inference for cosmological large scale structure!
arxiv.org/abs/2511.04681
arxiv.org/abs/2511.04681
November 10, 2025 at 1:37 AM
Exciting! Dark Energy Survey collaboration releases first paper demonstrating power of simulation-based inference for cosmological large scale structure!
arxiv.org/abs/2511.04681
arxiv.org/abs/2511.04681
I'm on my way to @caltech.edu for an AI + Science conference. Looking forward to seeing some friends and meeting new ones. There will be a livestream.
aiscienceconference.caltech.edu
aiscienceconference.caltech.edu
November 9, 2025 at 8:41 PM
I'm on my way to @caltech.edu for an AI + Science conference. Looking forward to seeing some friends and meeting new ones. There will be a livestream.
aiscienceconference.caltech.edu
aiscienceconference.caltech.edu
November 5, 2025 at 1:36 PM
For my @uwmadison.bsky.social colleagues… happy Halloween
November 1, 2025 at 1:14 AM
For my @uwmadison.bsky.social colleagues… happy Halloween
We also add a target variable θ∈ℝ that has known mutual information with the data sets x₁, ..., xₙ so that one can test how well a down-stream model with a shared representation can predict θ.
This is all aimed at making benchmark datasets and for powering ablation studies
This is all aimed at making benchmark datasets and for powering ablation studies
October 28, 2025 at 5:23 PM
We also add a target variable θ∈ℝ that has known mutual information with the data sets x₁, ..., xₙ so that one can test how well a down-stream model with a shared representation can predict θ.
This is all aimed at making benchmark datasets and for powering ablation studies
This is all aimed at making benchmark datasets and for powering ablation studies
We also use simple linear structural equations to tell a causal story with hyper-parameters that are easy to interpret. So we start with some "proto-latents" u that are totally uncorrelated and then construct the latents z through linear combinations. Here ũ₁ is a common cause
October 28, 2025 at 5:23 PM
We also use simple linear structural equations to tell a causal story with hyper-parameters that are easy to interpret. So we start with some "proto-latents" u that are totally uncorrelated and then construct the latents z through linear combinations. Here ũ₁ is a common cause
Each flow can be pre-trained to generate realistic looking data for a given setting... e.g. images, scientific data, etc. We used images for this demonstration.
We know the mutual information between this set of frogs and cars!
We know the mutual information between this set of frogs and cars!
October 28, 2025 at 5:23 PM
Each flow can be pre-trained to generate realistic looking data for a given setting... e.g. images, scientific data, etc. We used images for this demonstration.
We know the mutual information between this set of frogs and cars!
We know the mutual information between this set of frogs and cars!
The basic idea is to simultaneously generate latent variables z=(z₁, ..., zₙ) with known covariance (where each zᵢ ∈ ℝᵈ ) and then run each of them through an invertible transformation (flow): xᵢ = fᵢ(zᵢ)
That preserves the mutual information.
I(zᵢ, zⱼ) = I(xᵢ, xⱼ)
That preserves the mutual information.
I(zᵢ, zⱼ) = I(xᵢ, xⱼ)
October 28, 2025 at 5:23 PM
The basic idea is to simultaneously generate latent variables z=(z₁, ..., zₙ) with known covariance (where each zᵢ ∈ ℝᵈ ) and then run each of them through an invertible transformation (flow): xᵢ = fᵢ(zᵢ)
That preserves the mutual information.
I(zᵢ, zⱼ) = I(xᵢ, xⱼ)
That preserves the mutual information.
I(zᵢ, zⱼ) = I(xᵢ, xⱼ)
New paper, with @rkhashmani.me @marielpettee.bsky.social @garrettmerz.bsky.social Hellen Qu. We introduce a framework for generating realistic, highly multimodal datasets with explicitly calculable mutual information. This is helpful for studying self-supervised learning
arxiv.org/abs/2510.21686
arxiv.org/abs/2510.21686
October 28, 2025 at 5:23 PM
New paper, with @rkhashmani.me @marielpettee.bsky.social @garrettmerz.bsky.social Hellen Qu. We introduce a framework for generating realistic, highly multimodal datasets with explicitly calculable mutual information. This is helpful for studying self-supervised learning
arxiv.org/abs/2510.21686
arxiv.org/abs/2510.21686
It was a pleasure to meet all of our new colleagues and introduce them to our network of affiliates
October 23, 2025 at 5:01 PM
It was a pleasure to meet all of our new colleagues and introduce them to our network of affiliates
“Lies, damned lies, and next word prediction” — me, just now
October 13, 2025 at 8:24 PM
“Lies, damned lies, and next word prediction” — me, just now
Good morning moon @uwmadison.bsky.social
October 9, 2025 at 1:23 PM
Good morning moon @uwmadison.bsky.social
Ok, that’s cool. The Chicago Quantum Exchange has IBM quantum one in O’Hare next to the Brachiosaurus
October 7, 2025 at 1:06 AM
Ok, that’s cool. The Chicago Quantum Exchange has IBM quantum one in O’Hare next to the Brachiosaurus
I’m on my way to STAMPS @cmu.edu for a workshop on Simulation-Based Inference!
www.cmu.edu/dietrich/sta...
www.cmu.edu/dietrich/sta...
October 3, 2025 at 9:49 PM
I’m on my way to STAMPS @cmu.edu for a workshop on Simulation-Based Inference!
www.cmu.edu/dietrich/sta...
www.cmu.edu/dietrich/sta...
Is scale all you need? Or is there still a role for incorporating domain knowledge and inductive bias? While I was in Heidelberg, I took some time to write a short essay on this question called "The Bittersweet Lesson".
theoryandpractice.org/2025/09/The%... #HLF25
theoryandpractice.org/2025/09/The%... #HLF25
September 22, 2025 at 3:49 PM
Is scale all you need? Or is there still a role for incorporating domain knowledge and inductive bias? While I was in Heidelberg, I took some time to write a short essay on this question called "The Bittersweet Lesson".
theoryandpractice.org/2025/09/The%... #HLF25
theoryandpractice.org/2025/09/The%... #HLF25
Avi Wigderson gave a wonderful and inspiring talk about Turing at #HLF25. When he got to Turing’s theory of Morphogenesis he said
“When I see a problem like this, i throw up my arms and I go back to P vs NP, I’d have no idea how start” 🤣
“When I see a problem like this, i throw up my arms and I go back to P vs NP, I’d have no idea how start” 🤣
September 19, 2025 at 8:49 AM
Avi Wigderson gave a wonderful and inspiring talk about Turing at #HLF25. When he got to Turing’s theory of Morphogenesis he said
“When I see a problem like this, i throw up my arms and I go back to P vs NP, I’d have no idea how start” 🤣
“When I see a problem like this, i throw up my arms and I go back to P vs NP, I’d have no idea how start” 🤣
My office the last two mornings
September 18, 2025 at 9:36 AM
My office the last two mornings
An article in @physicsworld.bsky.social about our panel discussion on AI and the physical sciences at #HLF25. With
@georgemusser.com, Thea Åarrestad, David Silver and Maia Fraser
physicsworld.com/a/the-pros-a...
@georgemusser.com, Thea Åarrestad, David Silver and Maia Fraser
physicsworld.com/a/the-pros-a...
September 18, 2025 at 6:42 AM
An article in @physicsworld.bsky.social about our panel discussion on AI and the physical sciences at #HLF25. With
@georgemusser.com, Thea Åarrestad, David Silver and Maia Fraser
physicsworld.com/a/the-pros-a...
@georgemusser.com, Thea Åarrestad, David Silver and Maia Fraser
physicsworld.com/a/the-pros-a...
Happening soon: panel discussion on AI and the Physical Sciences at #HLF2025
www.heidelberg-laureate-forum.org/forum/12th-h...
www.heidelberg-laureate-forum.org/forum/12th-h...
September 16, 2025 at 12:03 PM
Happening soon: panel discussion on AI and the Physical Sciences at #HLF2025
www.heidelberg-laureate-forum.org/forum/12th-h...
www.heidelberg-laureate-forum.org/forum/12th-h...
The words going along with this slide are quite thoughtful.
More here:
www.incompleteideas.net/Talks/Albert...
More here:
www.incompleteideas.net/Talks/Albert...
September 15, 2025 at 8:22 AM
The words going along with this slide are quite thoughtful.
More here:
www.incompleteideas.net/Talks/Albert...
More here:
www.incompleteideas.net/Talks/Albert...
When life gives you lemons…
September 14, 2025 at 7:49 PM
When life gives you lemons…