Discrimination between the final state of t t¯ H and t t¯ b t¯ using neural network
Open Access
- 1 January 2021
- journal article
- research article
- Published by IOP Publishing in Journal of Physics: Conference Series
- Vol. 1725 (1), 012002
- https://doi.org/10.1088/1742-6596/1725/1/012002
Abstract
After the discovery of the Higgs boson in 2012 at the Large Hadron Collider (LHC), the effort to understand the detailed properties of the Higgs bosons started. Of particular importance is study of the Higgs coupling to the top quark. This coupling can be studied through the associated production of Higgs boson with top-antitop quark pair, t
H. This process however suffers from the indistinguishable background t
H
, since the Higgs boson decays predominately into bottom anti-bottom quark pair, b
. This study presents systematic approach of using machine learning (ML), specifically neural network method to distinguish between the process t
H and t
b
. Using input variables of kinematic variables (momentum), we found a signal efficiency of 46.7 % for signal events that have passed the preselection criteria. We conclude that the currently used input variables are not sufficient to discriminate between signal and background events, and we suggest that inclusion of input variables calculated from the fully reconstructed event could provide stronger discrimination between signal and background.
H. This process however suffers from the indistinguishable background tH, since the Higgs boson decays predominately into bottom anti-bottom quark pair, b. This study presents systematic approach of using machine learning (ML), specifically neural network method to distinguish between the process tH and tb. Using input variables of kinematic variables (momentum), we found a signal efficiency of 46.7 % for signal events that have passed the preselection criteria. We conclude that the currently used input variables are not sufficient to discriminate between signal and background events, and we suggest that inclusion of input variables calculated from the fully reconstructed event could provide stronger discrimination between signal and background.