The LHC Olympics 2020 a community challenge for anomaly detection in high energy physics

Abstract

A new paradigm for data-driven, model-agnostic new physics searches at colliders is emerging, and aims to leverage recent breakthroughs in anomaly detection and machine learning. In order to develop and benchmark new anomaly detection methods within this framework, it is essential to have standard datasets. To this end, we have created the LHC Olympics 2020, a community challenge accompanied by a set of simulated collider events. Participants in these Olympics have developed their methods using an R&D dataset and then tested them on black boxes: datasets with an unknown anomaly (or not). Methods made use of modern machine learning tools and were based on unsupervised learning (autoencoders, generative adversarial networks, normalizing flows), weakly supervised learning, and semi-supervised learning. This paper will review the LHC Olympics 2020 challenge, including an overview of the competition, a description of methods deployed in the competition, lessons learned from the experience, and implications for data analyses with future datasets as well as future colliders.

Autoencoders for semivisible jet detection

“…Here, an autoencoder-based analysis for SV jets detection will be presented. Unlike previous attempts [5][6][7][8][9][10] that use jet images (four-momenta of jet constituents), we incorporate high-level jet features and substructure variables computed based on the four-momenta of the jet constituents: Energy Flow Polynomials (EFPs) [11], Energy Correlation Functions (ECFs) [12] and their ratios: C 2 and D 2 , as well as the jet p T dispersion p T D [13] and jet axes [14]. These prove to be highly useful in discriminating between the QCD background and the SV jet signal [15].…”

Rare and Different: Anomaly Scores from a combination of likelihood and out-of-distribution models to detect new physics at the LHC

“…An increasingly popular method for finding such anomalous signals is to use anomaly detection techniques derived from Deep Learning (DL), see e.g. [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20].…”

“…In contrast to the unsupervised event classification methods proposed in this work, another type of anomaly may arise purely in the form of an overdensity in the bulk of the phase space that departs from the background model. Methods that search for such anomalies were explored in the LHC Olympics [20], and typically involve some form of DL-driven background estimation, as well as a density comparison between the data and the background to identify regions of interest. The method proposed here does not attempt to perform a density comparison, but instead aims to classify singular events as regular or anomalous with an anomaly score.…”

Deep Set Auto Encoders for Anomaly Detection in Particle Physics

“…The first is methods that use density estimation or look for bumps on smooth curves . The recent LHC Olympics [17] introduced four data sets of simulated hadronic LHC events. Many teams submitted results for new methods looking for hidden resonances in the data.…”