Speaker
Colin Carncrose Crovella
(University of Alabama (US))
Description
We present a Graph Neural Network-based approach to the problem of classifying jets as originating from either top quarks or the hadronization of a quark or gluon. We use high-fidelity simulated samples from the CMS Open Data, constructing graph nodes from the hits on the various subdetector layers. We compare the GNN performance with other algorithms. Finally, we discuss the technical benefits and challenges of the MATLAB implementation of this approach.
Author
Colin Carncrose Crovella
(University of Alabama (US))
Co-authors
Dr
Conor Daly
(The MathWorks, Inc.)
Gleyzer Sergei V
Dr
Samuel Somuyiwa
(The MathWorks, Inc.)
Shravan Chaudhari
Dr
Temo Vekua
(The MathWorks, Inc.)
