Working Group of Sascha Caron

Associate Professor

Radboud University and Nikhef

Artificial Intelligence and Data Science in Particle and Astrophysics



(Reload the webpage to restart the cellular automaton)

About

I lead a research group at the intersection of artificial intelligence and fundamental physics, pioneering new approaches to scientific discovery through machine learning. My work bridges experimental particle physics at CERN's ATLAS detector, theoretical physics, and astrophysics, with a focus on developing transformative AI methodologies for understanding dark matter and beyond-standard-model physics.

At the core of my current research is the development of Physical Data Encoders , Large Physics Models and Large Physics Benchmarks, a paradigm shift that brings the successes of foundation models from AI to fundamental physics. This vision combines decades of experience in particle physics data analysis with cutting-edge machine learning to create scalable, generalizable AI systems for physics discovery.

Beyond technical innovation, I've built and lead major international communities advancing AI in physics. As a founder of the DarkMachines Initiative and EuCAIF (European Coalition for AI in Fundamental Physics), I've created collaborative networks spanning over 100 researchers across Europe and beyond, establishing new frameworks for community-driven research at the frontier of AI and fundamental physics.

My research portfolio spans ATLAS collaboration leadership roles, pioneering work in generative AI for physics simulation and model-independent anomaly detection methods, and strategic contributions to European research policy through ECFA and the European Strategy for Particle Physics.

Advancing Discovery at the Energy Frontier

Research Highlights

EuCAIF Leadership

Established Europe's premier AI initiative for fundamental physics, officially recognized by ECFA, APPEC, and NuPECC. Named by the German Ministry (BMBF) in 2025 as a priority network for European AI strategy, building a community of over 60 senior fellows.

Explore EuCAIF

Large Physics Models & Benchmarks

Pioneering the conceptual framework for applying large-scale AI paradigms to fundamental physics. Introducing Large Physics Models as physics analogues to Large Language Models and developing standardized benchmarks for evaluation across collider and astrophysical data analysis.

Read Paper

DarkMachines Community

Founded the first global research initiative applying machine learning to dark matter discovery, building an international community of over 100 researchers. Delivered 6 community publications including the anomaly detection challenge that became a standard benchmark in the field.

Visit DarkMachines

First SUSY Search Results at LHC

Co-led first published supersymmetry searches at the LHC with over 1000 combined citations. Featured in Nature news coverage and excluded large regions of dark matter parameter space, establishing foundational methodologies for beyond-standard-model searches.

View Results

Model-Independent Discovery Framework

Pioneered general, data-driven anomaly detection in collider physics over a decade before modern AI techniques. Delivered the broadest systematic searches at HERA and LHC spanning over 700 channels and 10,000 signal regions.

Learn More

Generative AI for Physics Events

First publication using generative AI to simulate collider events, introducing the concept of an information buffer for VAEs. This work laid the foundation for fast simulation methods now widely adopted in particle physics.

Read Paper

ATLAS SUSY MET Coordinator Leadership

Co-convener of the ATLAS SUSY working group for missing transverse momentum searches, coordinating over 100 scientists during LHC's startup phase. Delivered the first official conference presentation of SUSY search results at Moriond 2011 (ATLAS experiment).

Cross-Experiment ML Innovation

Pioneered advanced ML and multivariate methods across H1, D0, and ATLAS experiments. Developed likelihood-based W-tagging, multivariate b-tagging for triggers, and transformer-based muon reconstruction, spanning two decades of innovation.

Interdisciplinary AI Applications

Built substantial independent research profile (H-index >20 outside collaborations) pioneering AI across theoretical physics, astrophysics, and scientific epistemology. Developed autosourceID framework and deep learning for gamma-ray astrophysics.

From Particle Physics to Cosmic Frontiers

Publications

Full publication list and citation metrics available on:

INSPIRE-HEP (All publications) INSPIRE-HEP (excluding the ATLAS collaboration) Google Scholar

Media & Outreach

Selected recent conference Talks

  • ACAT Workshop dedicated to Advanced Computing and Analysis Techniques in Physics, plenary/keynote presentation on “Foundation Models for Physics”, 2025
  • organized EuCAIFCon 2025 , Sardinia 2025.
  • CERN colloquium, Large Physics Models and EuCAIF, 2025
  • Presentation on AI in fundamental physics , ECFA plenary meeting, CERN, 2024
  • Member of Panel Discussion on Large Language Models for Physics, plenary, Krakau, CHEP Conference 2024
  • Examples of Press Coverage of our work

  • 2024: Volkskrant (NL) (“Komt de volgende ontdekking in de natuurkunde van AI?
  • 2018: ATLAS general searches, “LHC physicists embrace brute-force approach to particle hunt”, Nature news and reprinted in Scientific American
  • 2015: SUSY solutions Galactic Center Excess, “Mysterious galactic signal points LHC to dark matter”, Nature, Volume 521, Issue 7550, pp. 17-18 (2015).
  • Team

    Our research group consists of PhD students, postdocs, and collaborators working at the intersection of AI and fundamental physics.

    Polina Moskvitina

    PhD Student

    Research focus: Classification and Anomaly detection for 4 top events

    Nadezda Dobreva

    PhD Student

    Research focus: Developing transformer architectures for tracking and muon reconstruction

    Eugene Shalugin

    PhD Student

    Research focus: End to end di-higgs classification, reconstruction free higgsformers, Large Physics Benchmark

    Not assigned yet

    Post Doc

    Research focus: ML based Optimization of di-higgs analysis




    My former students and Postdocs:

    Contact

    Office

    FNWI (Huijgensgebouw), Radboud University
    Nijmegen, Netherlands

    Nikhef

    National Institute for Subatomic Physics
    Amsterdam, Netherlands

    Webpage is based on a template from w3schools, modified by me and Claude.