Parton shower uncertainties in jet substructure analyses with deep neural networks

Abstract

Machine learning methods incorporating deep neural networks have been the subject of recent proposals for new hadronic resonance taggers. These methods require training on a dataset produced by an event generator where the true class labels are known. However, this may bias the network towards learning features associated with the approximations to QCD used in that generator which are not present in real data. We therefore investigate the effects of variations in the modelling of the parton shower on the performance of deep neural network taggers using jet images from hadronic W-bosons at the LHC, including detector-related effects. By investigating network performance on samples from the Pythia, Herwig and Sherpa generators, we find differences of up to fifty percent in background rejection for fixed signal efficiency. We also introduce and study a method, which we dub zooming, for implementing scale-invariance in neural network-based taggers. We find that this leads to an improvement in performance across a wide range of jet transverse momenta. Our results emphasise the importance of gaining a detailed understanding of what aspects of jet physics these methods are exploiting. arXiv:1609.00607v2 [hep-ph]

Deep-learning top taggers or the end of QCD?

“…This distinguishes our approach from tagging methods which use Monte Carlo simulations for training, like the Template Tagger [9][10][11][12]. This means that for our performance test we do not have to include uncertainties in our Pythia simulations compared to other Monte Carlo simulations and data [51].…”

Jet tagging via particle clouds

“…The image-based approach has been extensively studied for various jet tagging tasks, e.g., W boson tagging [25][26][27][28][29]35], top tagging [32][33][34] and quark-gluon tagging [30,31]. Convolutional neural networks (CNNs) with various architectures were explored in these studies, and they were found to achieve sizable improvement in performance compared to traditional multivariate methods using observables motivated by QCD theory.…”

“…The implementation details can be found in Appendix A. Note that the ResNeXt-50 architecture is much deeper and therefore has a much larger capacity than most of the CNN architectures [25,[27][28][29][30][31][32][33][34][35] explored for jet tagging so far, so evaluating its performance on jet tagging will shed light on whether architectures for generic image classification is also applicable to jet images.…”

“…Recently, machine learning (ML) has injected fresh blood in jet tagging. Jets are regarded as images [25][26][27][28][29][30][31][32][33][34][35], as sequences [36][37][38][39][40][41][42][43][44][45][46][47], trees [48,49] or graphs [50] of particles, and ML techniques, most notably, deep neural networks (DNNs), are used to build new jet tagging algorithms automatically from (labelled) simulated samples or even (unlabelled) real data [51][52][53][54], leading to new insights and improvements in jet tagging.…”

Parton-shower uncertainties with Herwig 7: benchmarks at leading order

“…Recent development of these simulations has seen improvements in various areas, both within perturbative calculations, through matching to fixed order [2,[7][8][9][10][11][12][13][14][15], combining higher jet multiplicities [16][17][18][19][20][21][22], as well as the all-order resummation with parton showers [23][24][25][26] and also within the non-perturbative, phenomenological models [27,28]. While there are well established prescriptions on how to quantify the theoretical uncertainty of fixed-order calculations due to missing higher-order contributions [29][30][31][32][33][34][35] 1 , there is no such consensus for general resummed calculations [36][37][38][39][40][41], and parton-shower algorithms in particular [42][43][44][45][46][47], since a number of ambiguities are present within the different schemes; however, there is progress in towards this goal. Given the perturbative improvements, and the expected precision from data-taking at Run II of the Large Hadron Col- …”