
Johann Brehmer
@johannbrehmer
Machine learner & physicist. At @cusp_ai, I teach machines to discover materials for carbon capture. Previously Qualcomm AI Research, NYU, Heidelberg U.
ID: 1270423282966245378
https://johannbrehmer.github.io 09-06-2020 18:31:40
119 Tweet
1,1K Followers
536 Following




A milestone! 10 years ago I had an epiphany about using machine learning to approximate likelihood ratios, enabling statistical inference when your model is a complex simulator. Since then ~1000 papers on SBI have been published, but this is the 1st from the ATLAS Experiment đź§µ



Excited to share the first results using Neural Simulation-Based Inference (NSBI) techniques applied to ATLAS Experiment data! We measure the elusive off-shell Higgs boson with 3.1x better observation sensitivity than standard (histogram) analysis techniques! A thread: (1/N)


This new framework will extend the vision of the original "Madminer" papers by Kyle Cranmer , Johann Brehmer, Gilles Louppe and others to the LHC use case, potentially enabling wide-scale applications - accelerating discovery potential of the LHC (N/N)


Professional milestone: our review paper “The Frontier of Simulation-Based Inference” coauthored with Gilles Louppe and Johann Brehmer hit 1000 citations. I’m very excited about the potential for these methods to transform science! pnas.org/doi/10.1073/pn… simulation-based-inference.org




Congratulations to Johann Brehmer Pim de Haan Taco Cohen ! Thanks for saving equivariance.

ICLR 2025 MLMP best poster award goes to "ViNE-GATr: scaling geometric algebra transformers with virtual nodes embeddings"! Congratulations Julian Suk, Gabriele Cesa , Thomas Hehn, Arash Behboodi!
