In this paper we study the use of machine learning techniques to exploit kinematic information in $VH$, the production of a Higgs in association with a massive vector boson. We parametrize the effect of new physics in terms of the standard model effective field theory (SMEFT) framework. We find that the use of a shallow neural network allows us to dramatically increase the sensitivity to deviations in $VH$ respect to previous estimates. We also discuss the relation between the usual measures of performance in machine learning, such as area under the curve or accuracy, with the more adept measure of Asimov significance. This relation is particularly relevant when parametrizing systematic uncertainties. Our results show the potential of incorporating machine learning techniques to the SMEFT studies using the current datasets.

We study the phenomenological consequences of several CP-violating structures that could arise in the Standard Model effective field theory framework. Focusing on operators involving electroweak...

In this paper, a model inspired by Grand Unification principles featuring three generations of vector-like fermions, new Higgs doublets and a rich neutrino sector at the low scale is presented. Using the state-of-the-art Deep Learning techniques we perform the first phenomenological analysis of this model focusing on the study of new charged vector-like leptons (VLLs) and their possible signatures at CERN’s Large Hadron Collider (LHC). In our numerical analysis we consider signal events for vector-boson fusion and VLL pair production topologies, both involving a final state containing a pair of charged leptons of different flavor and two sterile neutrinos that provide a missing energy. We also consider the case of VLL single production where, in addition to a pair of sterile neutrinos, the final state contains only one charged lepton. We propose a novel method to identify missing transverse energy vectors by comparing the detector response with Monte-Carlo simulated data. All calculated observables are provided as data sets for Deep Learning analysis, where a neural network is constructed, based on results obtained via an evolutive algorithm, whose objective is to maximise either the accuracy metric or the Asimov significance for different masses of the VLL. Taking into account the effect of the three analysed topologies, we have found that the combined significance for the observation of new VLLs at the high-luminosity LHC can range from 5.7σ, for a mass of 1.25 TeV, all the way up to 28σif the VLL mass is 200 GeV. We have also shown that by the end of the LHC Run-III a 200 GeV VLL can be excluded with a confidence of 8.8 standard deviations. The results obtained show that our model can be probed well before the end of the LHC operations and, in particular, providing important phenomenological information to constrain the energy scale at which new gauge symmetries emergent from the considered Grand Unification picture can be manifest.

Currently, there is extensive research focused on automatic strategies for the segmentation and classification of 3D point clouds, which can accelerate the study of a landmark and integrate it with heterogeneous data and attributes, useful to facilitate the digital management of architectural heritage data. In this work, an automated image-based survey has been exploited a Region- Based Convolutional Neural Network. The training phase has been executed providing examples of images with the anomalies to be detected. At the same time, a laser scanning process was conducted to obtain a point cloud, which acts as a reference for the BIM process. In a final step, a process of projecting information from the images onto the BIM recreates the pathology shapes on the model’s objects, which generates a decision support system for the built environment. The innovation of this research concerns the development of a workflow in which it is possible to automatize the recognition and classification of defects in historical buildings, to finally interpolate this geometric and numerical information with a BIM methodology, obtaining a representation and quantification of the information adapted to the facility management process. The use of innovative techniques such as artificial intelligence algorithms and different plug-ins becomes the main strength of this project.

We demonstrate the application of the YOLOv5 model, a general purpose convolution-based object detection model, in the task of detecting binary neutron star coalescence events from gravitational-wave data of current generation interferometer detectors. We also present a thorough explanation of the synthetic data generation and preparation tasks based on approximant waveform models used for the model training, validation and testing steps. Using this approach, we achieve mean average precision [0.50] values of 0.945 for a single class validation dataset and as high as 0.978 for test datasets. Moreover, the trained model is successful in identifying the GW170817 event in the LIGO H1 detector data. The identification of this event is also possible for the LIGO L1 detector data with an additional preprocessing step, without the need of removing the large glitch in the final stages of the inspiral. The detection of the GW190425 event is less successful, which attests to performance degradation with the signal-to-noise ratio. Our study indicates that the YOLOv5 model is an interesting approach for first-stage detection alarm pipelines.

We propose a model with two Higgs doublets and several SU(2) scalar singlets with a global non-Abelian flavor symmetry $$ {\mathcal{Q}}_6\times {\mathcal{Z}}_2 $$Q 6 × Z 2 . This discrete group accounts for the observed pattern of fermion masses and mixing angles after spontaneous symmetry breaking. In this scenario only the third generation of fermions get their masses as in the Standard Model (SM). The masses of the remaining fermions are generated through a seesaw-like mechanism. To that end, the matter content of the model is enlarged by introducing electrically charged vector-like fermions (VLFs), right handed Majorana neutrinos and several SM scalar singlets. Here we study the processes involving VLFs that are within the reach of the Large Hadron Collider (LHC). We perform collider studies for vector-like leptons (VLLs) and vector-like quarks (VLQs), focusing on double production channels for both cases, while for VLLs single production topologies are also included. Utilizing genetic algorithms for neural network optimization, we determine the statistical significance for a hypothetical discovery at future LHC runs. In particular, we show that we can not safely exclude VLLs for masses greater than 200 GeV. For VLQ’s in our model, we show that we can probe their masses up to 3.8 TeV, if we take only into account the high-luminosity phase of the LHC. Considering Run-III luminosities, we can also exclude VLQs for masses up to 3.4 TeV. We also show how the model with predicted VLL masses accommodates the muon anomalous magnetic moment.