A Deep Neural Network for Simultaneous Estimation of b Jet Energy and Resolution

Search for Higgs Boson Decay to a Charm Quark-Antiquark Pair in Proton-Proton Collisions at sqrt[s]=13 TeV

A search for the standard model Higgs boson decaying to a charm quark-antiquark pair, H→cc[over ¯], produced in association with a leptonically decaying V (W or Z) boson is presented. The search is performed with proton-proton collisions at sqrt[s]=13  TeV collected by the CMS experiment, corresponding to an integrated luminosity of 138  fb^{-1}. Novel charm jet identification and analysis methods using machine learning techniques are employed. The analysis is validated by searching for Z→cc[over ¯] in VZ events, leading to its first observation at a hadron collider with a significance of 5.7 standard deviations. The observed (expected) upper limit on σ(VH)B(H→cc[over ¯]) is 0.94 (0.50_{-0.15}^{+0.22})pb at 95% confidence level (C.L.), corresponding to 14 (7.6_{-2.3}^{+3.4}) times the standard model prediction. For the Higgs-charm Yukawa coupling modifier, κ_{c}, the observed (expected) 95% C.L. interval is 1.1<|κ_{c}|<5.5 (|κ_{c}|<3.4), the most stringent constraint to date.

Search for Higgs Boson Pair Production in the Four b Quark Final State in Proton-Proton Collisions at sqrt[s]=13 TeV

A search for pairs of Higgs bosons produced via gluon and vector boson fusion is presented, focusing on the four b quark final state. The data sample consists of proton-proton collisions at a center-of-mass energy of 13 TeV, collected with the CMS detector at the LHC, and corresponds to an integrated luminosity of 138  fb^{-1}. No deviation from the background-only hypothesis is observed. A 95% confidence level upper limit on the Higgs boson pair production cross section is observed at 3.9 times the standard model prediction for an expected value of 7.8. Constraints are also set on the modifiers of the Higgs field self-coupling, κ_{λ}, and of the coupling of two Higgs bosons to two vector bosons, κ_{2 V}. The observed (expected) allowed intervals at the 95% confidence level are -2.3<κ_{λ}<9.4 (-5.0<κ_{λ}<12.0) and -0.1<κ_{2 V}<2.2 (-0.4<κ_{2 V}<2.5). These are the most stringent observed constraints to date on the HH production cross section and on the κ_{2 V } coupling.

Learning Uncertainties the Frequentist Way: Calibration and Correlation in High Energy Physics

Calibration is a common experimental physics problem, whose goal is to infer the value and uncertainty of an unobservable quantity Z given a measured quantity X. Additionally, one would like to quantify the extent to which X and Z are correlated. In this Letter, we present a machine learning framework for performing frequentist maximum likelihood inference with Gaussian uncertainty estimation, which also quantifies the mutual information between the unobservable and measured quantities. This framework uses the Donsker-Varadhan representation of the Kullback-Leibler divergence-parametrized with a novel Gaussian ansatz-to enable a simultaneous extraction of the maximum likelihood values, uncertainties, and mutual information in a single training. We demonstrate our framework by extracting jet energy corrections and resolution factors from a simulation of the CMS detector at the Large Hadron Collider. By leveraging the high-dimensional feature space inside jets, we improve upon the nominal CMS jet resolution by upward of 15%.

Lightweight jet reconstruction and identification as an object detection task

We apply object detection techniques based on deep convolutional blocks to end-to-end jet identification and reconstruction tasks encountered at the CERN large hadron collider (LHC). Collision events produced at the LHC and represented as an image composed of calorimeter and tracker cells are given as an input to a Single Shot Detection network. The algorithm, named PFJet-SSD performs simultaneous localization, classification and regression tasks to cluster jets and reconstruct their features. This all-in-one single feed-forward pass gives advantages in terms of execution time and an improved accuracy w.r.t. traditional rule-based methods. A further gain is obtained from network slimming, homogeneous quantization, and optimized runtime for meeting memory and latency constraints of a typical real-time processing environment. We experiment with 8-bit and ternary quantization, benchmarking their accuracy and inference latency against a single-precision floating-point. We show that the ternary network closely matches the performance of its full-precision equivalent and outperforms the state-of-the-art rule-based algorithm. Finally, we report the inference latency on different hardware platforms and discuss future applications.