Who am I?

I am a particle physicst currently working as a senior research fellow at ETH Zurich. Here I work on utilizing Machine Learning to improve the way we collect and analyze data at the Large Hadron Collider. I have a PhD in physics from the University of Zurich, where I worked on searches for new heavy particles decaying into dibosons and jet substructure techniques.

Currently, I specialize in real-time Machine Learning to help us process enormous data rates (63 Tb/s) at LHC, within a latency of 12 microseconds. I also work on utilizing ML-based anomaly detection as a way of discovering New Physics as data outliers. When I am not writing code or trail running, I supervise students on various projects as well as give schools and seminars on ML in partice physics.

As any particle physicist at CERN, I spend most of my days being a combination of a data scientist, statistician, experimental physicist, teacher and software engineer. CERN is a vibrating and inspiring community, where I daily work together with people from all around the globe who are experts in their field. If you want to hear more, feel free to drop me an email below!

Recent news

AMD Tech Summit Keynote

It was a pleasure to visit AMD in Dublin and give a keynote on real-time inference in particle physics experiments!

Paper out! Anomaly Detection for New Physics Searches

The pre-print of our paper on ML-accelerated anomaly detection for New Physics searches is now out!

Symmetry Magazine Interview

On why I believe collaborations with industry are essential for the future of particle physics.

Nature interview: CERNs impact

Had a chat with Nature about CERN's impact on society!

IEEE Nuclear Science Symposium keynote

Honored to be a plenary speaker at IEEE Nuclear Science Symposium 2024 in Florida! Also convening the track AI and Machine Learning for Radiation Detection together with Audrey Corbeil Therrien, and I am excited to see the many applications of AI in nuclear science!

Generative AI in the Physical Sciences

Look forward for talking about Physics-Motivated Approaches to Hardware Design (distributed foundation models on FPGAs) at MIT next week!

Seminar at Rutherford Appleton Laboratory

Really enjoyed visiting RAL, an impressive laboratory with a vast research program. A recording of my seminar is available above!

Group photo fall semester 2023

This semester I was lucky to have four fantastic students: Kyle Metzger (Physics), Matthias Bonvin (Applied Maths), Patrick Odagiu (Physics, PhD) and Jessica Prendi (Physics). I am very happy that I get to keep all of them as master students next semester!

ETH IPAi visits CMS

My students Jessica Prendi and Patrick Odagiu got to visit CMS today during the winter shutdown

Google visits CERN

CERN with our collaborators Richard Stotz and Mathieu Guillame-Bert from Google Zurich.

Norwegian Teacher's Program

I brought my daughter to CERN to teach Norwegian highschool teachers about Machine Learning and particle physics. The CERN teacher program is fantastic!

Coffee lecture for ETH students

It was an honour to be invited to talk to mathematics, physics and computer science students at ETH about how to make an academic career in physics as a woman.

Anomaly Detection in Particle Physics

Our review paper on anomaly detection for particle physics has appeared in Review in Physics. A comprehensive overview on anomaly detection for New Physics searches and for triggering!

ML at L1T Workshop at CERN

We had a very productive workshop dedicated to ML algorithms for the CMS Level-1 hardware trigger. Sioni Summers, Artur Lobanov and I put together a nice tutorial on how to design, compress and deploy ML algorithms on the L1T FPGAs. Recordings available above!

Hammers & Nails 2023

Hammers & Nails is a great workshop focused on how we can use the machine learning hammer to do better science. With open-ended lectures spanning academia and industry, it is a place where ideas are born. I gave an invited talk about real-time machine learning on specialized hardware, that you can find on the link above.

Fast Machine Learning for Science 2023

Our yearly Fast ML Workshop was at Imperial College in London this year. It featured an excellent program with speakers from biomedicine, free-electron laser accelerators, laser wakefields, fusion, accelerator control and much more. I was excited to give an overiew talk over fast ML at the LHC.

Maria Laach school: ML lecture series

It was an honour getting to lecture at the renowned Herbstschule fur Hochenergiephysik Maria Laach. I gave a three hour lecture series on Machine Learning in Particle Physics.

AI in the Physical Sciences

Tobias Golling (University of Geneva) and I had a very interesting discussion on how we to leverage AI in the Physical Sciences with our panelists Francois Charton (META Paris), Francois Fleuret (University of Geneva), Michael Kagan (SLAC), and Sofia Vallecorsa (CERN) at AI2S2 in Geneva today!

My MLSS^S lecture is now online!

My lecture on ML at CERN for the ML Summer School in Krakow is now on YouTube!

AXOL1TL

How do you like our project mascot? Made by Noah Zipper

CVPR 2023: Real-time inference for event cameras

Our paper on within camera enoising for DVS cameras was accepted for CVPR 2023. It was a great pleasure to collaborate with other departments and universities on something I do not usually work on.

A3D3 Seminar: Leveraging Real-Time ML for Handling Massive LHC Data Streams

My A3D3 seminar is now on YouTube!

CERN Data Science Seminar: Real-time ML in Particle Physics

Over 500 people followed my talk at CERN on our work on real-time inference in particle physics! The talk is now online

Machine Learning Summer School^Science

FLooking forward for joining this line-up of excellent speakers in Krakow!

Our collaboration with Volvo's Zenseact

We had a lot of fun working with Zenseact on real-time inference for self-driving cars!

Our Google-CERN collaboration in the news!

Our collaboration with Google highlighted in the CERN Courier

Applied ML Days: Women in ML

Honored to be promoted amongst these talented women doing Machine Learning!

CHIPP Prize: Top 3 Particle Physics PhD theses in Switzerland

My PhD thesis was nominated for the 2019 CHIPP Prize! The purpose is to reward the best PhD student in Experimental or Theoretical Particle Physics and I was nominated together with two other Swiss PhD candidates. Unfortunately I did not win, but I am very proud to have been nominated together with the two other excellent candidates. Find my thesis summary that got nominated attached!

Finally a doctor!

I am a doctor! I very much enjoyed presenting my PhD thesis to friends, family and colleagues. You can find my presentation here!

ML4HEP: Evening lectures

In connection with the ML4HEP school, have a look at these two great evening talks we are organizing!

How to do ultrafast Deep Neural Network inference on FPGAs

Sign up for our one-day course on ultrafast DNN inference on FPGAs!

Machine Learning for High Energy Physics

Check out the Machine Learning school I'm organizing at UZH here!

Searches for new VV resonances

This years overview talk on diboson resonances from the BOOST conference

LoLa: Lorentz Invariance in TensorFlow

My poster on LoLa: A Lorentz Invariance Based DNN for W-tagging

Lorentz Invariance Based DNN for W-tagging

A talk I gave on a Lorentz Invariance Based DNN for discrimination

Search for massive resonances decaying into WW,WZ or ZZ bosons in proton-proton collisions at 13 TeV.

Our paper on the first 13 TeV search for diboson resonances in the all-hadronic final state with the CMS detector.

Search for massive resonances decaying into WW, WZ, ZZ, qW, and qZ with dijet final states at 13 TeV.

Our paper on searches for diboson resonances with a novel W-tagger using PUPPI+softdrop.

W/Z/H tagging in CMS

A talk I gave on jet substructure at BOOST 2017

Search for massive resonances decaying into WW, WZ, ZZ, qW, and qZ with dijet final states at 13 TeV.

Our paper on searches for diboson resonances with a novel W-tagger using PUPPI+softdrop.

Algorithms to identify high-energy B hadrons via their hit multiplicity increase through pixel detection layers.

Our student Manuels talk (https://indico.cern.ch/event/759973/, restricted) and thesis on the nice DNN b-tagger he developed for his Bachelor Thesis

Get In Touch

Feel free to drop me a message below!

  • Address

    Institute for Particle Physics
    Otto-Stern-Weg 5
    8093 Zürich
    Office HPK E 29
    Switzerland
  • Phone

    +41 44 633 45 68
  • Email

    thea.aarrestad@cern.ch