Machine learning in the search for new fundamental physics

  title={Machine learning in the search for new fundamental physics},
  author={Georgia Karagiorgi and Gregor Kasieczka and S. Kravitz and Benjamin Philip Nachman and David Shih},
  journal={Nature Reviews Physics},
Georgia Karagiorgi,1, ∗ Gregor Kasieczka,2, † Scott Kravitz,3, ‡ Benjamin Nachman,3, 4, § and David Shih5, ¶ 1Department of Physics, Columbia University, New York, NY 10027, USA 2Institut für Experimentalphysik, Universität Hamburg, 22761 Hamburg, Germany 3Physics Division, Lawrence Berkeley National Laboratory, Berkeley, CA 94720, USA 4Berkeley Institute for Data Science, University of California, Berkeley, CA 94720, USA 5NHETC, Department of Physics and Astronomy, Rutgers University… 

Jet Physics from the Ground Up

These are lecture notes presented at the online 2021 QUC Winter School on Energy Frontier hosted by the Korea Institute for Advanced Study. They extend lectures presented at the 2017 and 2018 CTEQ

Riemannian Data preprocessing in Machine Learning to focus on QCD color structure

Identifying the quantum chromodynamics (QCD) color structure of processes provides additional information to enhance the reach for new physics searches at the Large Hadron Collider (LHC). Analyses of

On scientific understanding with artificial intelligence

An oracle that correctly predicts the outcome of every particle physics experiment, the products of every possible chemical reaction or the function of every protein would revolutionize science and

DeeLeMa: Missing information search with Deep Learning for Mass estimation

We present DeeLeMa , a deep learning network to analyze energies and momenta in particle collisions at high energy colliders, especially for collision events with multiple invisible particles.

Self-supervised anomaly detection for new physics

A transformer neural network is trained to encode simulated QCD “event space” dijets into a low-dimensional “latent space’s” representation, which opens the door to usingLow-dimensional latent representations as a computationally efficient space for resonant anomaly detection in generic particle collision events.

Efficiently Moving Instead of Reweighting Collider Events with Machine Learning

There are many cases in collider physics and elsewhere where a calibration dataset is used to predict the known physics and / or noise of a target region of phase space. This calibration dataset

Data Science and Machine Learning in Education

Topics covered include handling, visualizing and adapting structure in data, adapting linear methods to nonlinear problems, density esti-mation, Bayesian statistics, Markov-chain Monte Carlo, variational inference, probabilistic programming,Bayesian model selection, artificial neural networks and deep learning.

Jets and Jet Substructure at Future Colliders

Even though jet substructure was not an original design consideration for the Large Hadron Collider (LHC) experiments, it has emerged as an essential tool for the current physics program. We examine

Anomaly Detection with Multiple Reference Datasets

This work proposes generalizations of CW O L A and SALAD for settings where multiple reference datasets are available, building on weak supervision techniques and demonstrates improved performance in a number of settings with real and synthetic data.

Benchmarking energy consumption and latency for neuromorphic computing in condensed matter and particle physics

A methodology for measuring the energy cost and compute time for inference tasks with ANNs on conventional hardware and an architecture for these tasks are designed and the same metrics are estimated based on a state-of-the-art analog in-memory computing platform, one of the key paradigms in neuromorphic computing.



Preserving new physics while simultaneously unfolding all observables

Patrick Komiske,1, 2, ∗ W. Patrick McCormack,3, 4, † and Benjamin Nachman4, 5, ‡ 1Center for Theoretical Physics, Massachusetts Institute of Technology, Cambridge, MA 02139, USA 2The NSF AI Institute

Anomaly Detection for Resonant New Physics with Machine Learning.

A new model-agnostic anomaly detection technique that naturally benefits from modern machine learning algorithms is presented that can turn a modest 2σ excess into a 7σ excess for a model with an intermediate BSM particle that is not currently targeted by a dedicated search.

Learning multivariate new physics

We discuss a method that employs a multilayer perceptron to detect deviations from a reference model in large multivariate datasets. Our data analysis strategy does not rely on any prior assumption

Learning new physics from a machine

This work proposes using neural networks to detect data departures from a given reference model, with no prior bias on the nature of the new physics responsible for the discrepancy, and constructs an algorithm that implements this approach, as a straightforward application of the likelihood-ratio hypothesis test.

Variational autoencoders for new physics mining at the Large Hadron Collider

A one-sided threshold test to isolate previously unseen processes as outlier events is developed, which could inspire new-physics model building and new experimental searches, and be complementary to classic LHC searches.

Quasi anomalous knowledge: searching for new physics with embedded knowledge

This paper presents a new strategy dubbed Quasi Anomalous Knowledge (QUAK), whereby it introduces alternative signal priors that capture some of the salient features of new physics signatures, allowing for the recovery of sensitivity even when the alternative signal is incorrect.

Machine learning at the energy and intensity frontiers of particle physics

The application and development of machine-learning methods used in experiments at the frontiers of particle physics (such as the Large Hadron Collider) are reviewed, including recent advances based on deep learning.

Neural Message Passing for Jet Physics

This work considers message passing on a graph where the nodes are the particles in a jet, and designs variants of a message-passing neural network with a learnable adjacency matrix and two message-Passing iterations that outperforms all the others.