Finding gluon jets with a neural trigger

Lönnblad, Leif; Peterson, Carsten; Rögnvaldsson, Thorsteinn

doi:10.1103/physrevlett.65.1321

Cited by 86 publications

(45 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Neural networks have been applied to a wide variety of problems in high-energy physics [1,2], from event classification [3,4] to object reconstruction [5,6] and triggering [7,8]. Typically, however, these networks are applied to solve a specific isolated problem, even when this problem is part of a set of closely related problems.…”

Section: Introductionmentioning

confidence: 99%

Parameterized neural networks for high-energy physics

et al. 2016

View full text Add to dashboard Cite

We investigate a new structure for machine learning classifiers built with neural networks and applied to problems in high-energy physics by expanding the inputs to include not only measured features but also physics parameters. The physics parameters represent a smoothly varying learning task, and the resulting parameterized classifier can smoothly interpolate between them and replace sets of classifiers trained at individual values. This simplifies the training process and gives improved performance at intermediate values, even for complex problems requiring deep learning. Applications include tools parameterized in terms of theoretical model parameters, such as the mass of a particle, which allow for a single network to provide improved discrimination across a range of masses. This concept is simple to implement and allows for optimized interpolatable results.

show abstract

Section: Introductionmentioning

confidence: 99%

Parameterized neural networks for high-energy physics

et al. 2016

View full text Add to dashboard Cite

show abstract

“…these methods in the early 1990s when they were brought into HEP analyses (19)(20)(21)(22)(23). However, following several successful applications (24)(25)(26)(27)(28)(29), particle physicists have largely accepted the use of NNs and other multivariate methods.…”

Section: P a G E P C B H A Tmentioning

confidence: 99%

Multivariate Analysis Methods in Particle Physics

Bhat

2011

Annu. Rev. Nucl. Part. Sci.

View full text Add to dashboard Cite

"That is positively the dopiest idea I have heard." -Richard Feynman, when he signed on to work on the Connection Machine, at the Thinking Machines Corporation, in the summer of 1983. Key WordsMultivariate methods, optimal analysis, neural networks, Bayesian inference, Tevatron, Large Hadron Collider (LHC) AbstractEach generation of high energy physics experiments is grander in scale than the previousmore powerful, more complex and more demanding in terms of data handling and analysis.The spectacular performance of the Tevatron and the beginning of operations of the Large Hadron Collider have placed us at the threshold of a new era in particle physics. The discovery of the Higgs boson or another agent of electroweak symmetry breaking and evidence of new physics may be just around the corner. The greatest challenge in these pursuits is to extract the extremely rare signals, if any, from huge backgrounds that arise from known physics processes. The use of advanced analysis techniques is crucial in achieving this goal. In this review, I discuss the concepts of optimal analysis, some important advanced analysis methods and a few examples. The judicious use of these advanced methods should enable new discoveries and produce results with better precision, robustness and clarity.

show abstract

“…Neural networks have proven to be an efficient technique for pattern recognition with important applications in a variety of high-energy physics problems [8]; our application is another in the area of signal-versusbackground discrimination.…”

mentioning

confidence: 99%

Snagging the top quark with a neural net

Baer¹,

Karatas²,

Giudice³

1992

Phys. Rev. D

View full text Add to dashboard Cite

The search for the top quark at pjj colliders in the one-lepton-plus-jets channel is plagued by an irremovable background from W-boson-plus-multijet production. In this paper, we show how the top-quark signal can be distinguished from background in the distribution of neural network output. By making a cut on the network output, we maximize the ratio of signal to background in a final event sample, and compare our results with those obtained by making kinematical cuts on the data sample. We also demonstrate the robustness of the neural network method by training the neural network on signal events of one top mass and testing upon another. PACS number(s): 13.85.Qk, 14.80.DqThe top quark t is a necessary ingredient of the standard model (SM), as it is required to ensure the cancellation of triangle anomalies [I], to account for the e+e-+ b6 forward-backward jet asymmetry 121, and for the suppression of neutral current B decays [3]. Radiative corrections to SM parameters measured at the CERN e+e-collider LEP constrain the top-quark mass to be 89 GeV < mt < 191 GeV 141. The lack of a topquark signal at the Fermilab Tevatron to date furthermore restricts mt > 91 GeV 151. The search for top quarks of mass up to -120 -140 GeV at the Fermilab Tevatron is expected to resume in 1992, with the planned collection of 20 pb-l of data.The top quark at the Fermilab Tevatron collider is expected to be produced dominantly via the gg --+ tf and qq --+ tf pair production channels. Since mt > Mw + ma, a SM top quark decays via t + bW where the W is on shell; hence, a ti? event results in a b6W+Wd final state. The cleanest signal for tf occurs in the two-lepton channel; the two leptons arise from the leptonic decay of each W boson. The principal backgrounds from Wpair production and y, Z --, T+T-are expected to be controllable [6]. Event rates in the twelepton channel will be low because of the two leptonic branching ratios; moreover, this channel does not provide a direct determination of the topquark mass because the two neutrinos from W + lv carry off an unknown amount of pr. The best one can do is to either match the total event rate to QCD predictions (which depend on mt), or to measure the relative rates, Rllo and Rzll, of multijet production in events with two isolated leptons 161. Here, Rilj stands for the ratio of events with i jets to events with j jets.The single-lepton channel of tf decay, on the other hand, in which one W decays leptonically and the other hadronically, offers the possibility of a direct determination of the topquark mass, since the presence of only a single neutrino permits a kinematical reconstruction of the event. By computing the invariant mass of the three leading jets for each event, the first two of which are expected to come from the hadronic W and the third from one of two b quarks, one may be able to determine the top-quark mass. This channel will be sought by the experimentalists both for its confirmation of the existence of the top quark, as well as for its value in providing a direct reconstru...

show abstract

Finding gluon jets with a neural trigger

Cited by 86 publications

References 9 publications

Parameterized neural networks for high-energy physics

Parameterized neural networks for high-energy physics

Multivariate Analysis Methods in Particle Physics

Snagging the top quark with a neural net

Contact Info

Product

Resources

About