The grow in power of computers and the enormous amount of data available from experiments lead in the last years to the birth of new methodologies to treat data analysis. Artificial Intelligence (AI) techniques such as neural network are now commonly used by scientist all over the world but, although wide spread, are currently not fully understood. This situation requires a study of the proprieties of the algorithm used in each specific case. In this thesis I tried to look after the techniques used in the analysis of the Bs->Ds*pi decay, using data from the LHCb experiment operated at the LHC, CERN in order to investigate an enhancement of the performance of the classification of the events. The Bs->Ds*pi channel is used as control for the Bs->KDs* channel which is hoped to be useful for a new/independent measurement of the angle gamma from the Cabibbo Kobayashi Maskawa matrix. The large amount of variables that define this decay process make it difficult to separate Background from Signal because both populate complicated regions in a large multidimensional space. The AI approach is mandatory to find where the main part of the Signal lies. To obtain these results an AI technique is trained with a mixture of real data and Monte-Carlo simulations and, once fully trained, the algorithm can be applied to additional data to enhance the Signal component. My first objective in this thesis was to take the actual ROOT based machine learning algorithm used at LHCb and try to boost its performances by changing how the training is done; later I tried to compare different ROOT based methods to see which one had the best performances. Last I tried to compare the ROOT based methods with an SKLearn based application to find out if more recent algorithms have better performances.

Study of the production of Bs mesons in the Ds*pi decay channel using LHCB data

Favaro, Giulio
2017/2018

Abstract

The grow in power of computers and the enormous amount of data available from experiments lead in the last years to the birth of new methodologies to treat data analysis. Artificial Intelligence (AI) techniques such as neural network are now commonly used by scientist all over the world but, although wide spread, are currently not fully understood. This situation requires a study of the proprieties of the algorithm used in each specific case. In this thesis I tried to look after the techniques used in the analysis of the Bs->Ds*pi decay, using data from the LHCb experiment operated at the LHC, CERN in order to investigate an enhancement of the performance of the classification of the events. The Bs->Ds*pi channel is used as control for the Bs->KDs* channel which is hoped to be useful for a new/independent measurement of the angle gamma from the Cabibbo Kobayashi Maskawa matrix. The large amount of variables that define this decay process make it difficult to separate Background from Signal because both populate complicated regions in a large multidimensional space. The AI approach is mandatory to find where the main part of the Signal lies. To obtain these results an AI technique is trained with a mixture of real data and Monte-Carlo simulations and, once fully trained, the algorithm can be applied to additional data to enhance the Signal component. My first objective in this thesis was to take the actual ROOT based machine learning algorithm used at LHCb and try to boost its performances by changing how the training is done; later I tried to compare different ROOT based methods to see which one had the best performances. Last I tried to compare the ROOT based methods with an SKLearn based application to find out if more recent algorithms have better performances.
2017-04
20
Machine learning, ROOT, Data analysis, Artificial Intelligence, LHC, CERN, LHCb
File in questo prodotto:
File Dimensione Formato  
Tesi_L_Favaro_Giulio.pdf

accesso aperto

Dimensione 505.99 kB
Formato Adobe PDF
505.99 kB Adobe PDF Visualizza/Apri

The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/20.500.12608/26253