In this talk we will present a new algorithm to search for new physics called Anomaly Awareness. By making our algorithm 'aware' of the presence of a range of different anomalies, we improve its capability to detect anomalous events even when it hasn't been exposed to them in the past. As an example, we apply this method to boosted jets and use it to uncover new resonances or EFT effects.
Higgs Bosons produced via gluon-gluon fusion (ggF) with large transverse momentum ($p_T$) are sensitive probes of physics Beyond the Standard Model. However, high $p_T$ Higgs Boson production is contaminated by a diversity of production modes other than ggF: vector boson fusion, production of a Higgs boson in association with a vector boson and with a top-quark pair. Combining jet substructure...
A new approach for the identification of VBF topology is presented. A Recurrent Neural Network (RNN) approach based on the 4-momentum of the small-R jets in the event has been developed in the context of the search of high mass resonances decaying into diboson semi-leptonic final states (X โ> VV โ> vv/lv/ll + qq). The class of RNN networks shows high performances and opportunity to deal with...
Entanglement is a key subject in quantum information theory. Due to its genuine relativistic and fundamental nature, high-energy colliders are attractive systems for the experimental study of quantum information theory. We propose the detection of entanglement between the spins of top-antitop quark pairs at the LHC, representing the first proposal of entanglement detection in a pair of quarks,...
We introduce a new event shape observable -- event isotropy -- that quantifies how close the radiation pattern of a collider event is to a uniform distribution. This observable is based on a normalized version of the energy mover's distance, which is the minimum "work" needed to rearrange one radiation pattern into another of equal energy. We investigate the utility of event isotropy both at...
High-precision all-order calculations can only be performed for a narrow class of observables, which are sensitive to radiation over the entire final state phase-space. When phase-space boundaries are introduced, the resummation is affected by so-called non-global logarithms, which have an intricate all-order structure. In this talk, we present a first-principle calculation for the non-global...
We study the link between parton dynamics in the collinear limit and the logarithmically enhanced terms of the groomed jet mass distribution, for jets groomed with the modified mass-drop tagger (mMDT). While the leading logarithmic structure is linked to collinear evolution with leading-order splitting kernels, here we derive the NLL structure directly from triple-collinear splitting...
A measurement of the production cross section for high transverse momentum top quark pairs is reported. The data set was collected during 2016 with the CMS detector at the LHC from pp collisions at 13 TeV, and corresponds to an integrated luminosity of 35.9 fb-1. The measurement uses events where either both top quark candidates decay hadronically and are reconstructed as large-radius jets...
First results are presented on the use of a new machine-learning based unfolding technique, OmniFold, applied to archival hadronic e+e- collisions using 730 pb^-1 of data collected at 91 GeV with the ALEPH detector at LEP. With the archived data and unfolding procedure, multiple classic hadronic event-shape variables are measured in a fully unbinned and multi-differential manner. Of particular...
I will present the search for boosted Higgs boson with transverse momentum greater than 450 GeV decaying into bottom quark pairs using LHC full run 2 dataset collected by the CMS experiment.
In this search, we employed the latest jet substructure variables and b-tagging techniques based on a deep neural network to reduce the overwhelming QCD backgrounds.
An excess of events above background...
The past few years have seen a rapid development of machine-learning algorithms. While surely augmenting performance, these complex tools are often treated as black-boxes and may impair our understanding of the physical processes under study. Moving a first step into the direction of applying expert-knowledge in particle physics, we test whether the optimal decision function is achieved by...
Measurements of the jet substructure in Pb+Pb collisions provide information about the mechanism of jet quenching in the hot and dense QCD medium created in these collisions, over a wide range of energy scales. This poster presents the ATLAS measurement of the suppression of yields of large-radius jets and its dependence on the jet substructure, characterized by the presence of sub-jets and...
A method is presented to extract salient information from a deep neural network classifier of jet substructure tagging techniques, using expert variables that augment the inputs, using layerwise relevance propagation. The results show that these eXpert AUGmented (XAUG) variables can be used to easily interpret the behavior of the classifier, and in some cases can capture the behavior of the...
Separating charged and neutral pions as well as calibrating the pion energy response is a core component of reconstruction in the ATLAS calorimeter. This poster presents an investigation of deep learning techniques for these tasks, representing the signal in the ATLAS calorimeter layers as pixelated images. Machine learning approaches outperform the classification applied in the baseline local...
Jet taggers that are decorrelated from certain observables, such as mass, are of increasing interest for experimental measurements.
Several methods have been proposed to design taggers that balance discrimination power against correlation.
As a fundamentally multi-objective optimization problem, there is an infinite set of Pareto-efficient solutions, known as the Pareto frontier.
We...
The reconstruction and identification of boosted hadronic final states is a key part for beyond the Standard Model (BSM) physics searches and precision measurements of Standard Model processes at ATLAS. Identification algorithms designed to identify boosted hadronically decaying W bosons and top quarks, known as taggers, have been updated and optimized from previous efforts to include the data...
The Monte Carlo simulation of calorimeter showers is a vital part of particle physics. However, individually modeling the paths and interactions of each particle in a shower is a very time consuming process. This computation time requirement becomes even more problematic as we move to higher luminosities. Therefore we aim to speed up shower simulations through the use of Generative Machine...
