July 27, 2021 Master Thesis Open Access

Sowa, Lars

Citation Style Language JSON Export

{
  "abstract": "<p>The SuperKEKB collider in Japan accelerates<br>\nelectrons and positrons which collide at an energy of 11 GeV aiming to produce B meson<br>\npairs. To analyse the properties of a B meson, it gets recombined from its decay daughters.<br>\nSince there are a lot of background events that mimic the signature of a B meson, it is<br>\nimportant to suppress these events. Such a suppression is called Continuum Suppression. This work aims to further improve the Deep Continuum Suppression, which is perfromed by a Multilayer Perceptron (MLP), by three points: Firstly, an MLP needs a fixed order for input particles. Therefore, this thesis<br>\npresents a reliable, self-attention-based input mechanism which allows for invariance under<br>\nthe particle order. Secondly, to guarantee certainty about the prediction of deep learning models, there is a<br>\nhigh interest in models with predictive uncertainties. This is addressed using the concept<br>\nof Deep Ensembles to predict uncertainties of continuum classifications.<br>\nFinally, the use of vertex information for the training of a model leads to a bias in<br>\ncertain analysis variables. Since this could lead to falsification of further studies, a Distance<br>\nCorrelation is used to decorrelate the Continuum Suppression model from third variables.</p>", 
  "author": [
    {
      "family": "Sowa, Lars"
    }
  ], 
  "id": "22060", 
  "issued": {
    "date-parts": [
      [
        2021, 
        7, 
        27
      ]
    ]
  }, 
  "language": "eng", 
  "title": "Deep Continuum Suppression with Predictive Uncertainties at the Belle II Experiment", 
  "type": "thesis"
}