Bonn 2020 – scientific programme

The DPG Spring Meeting in Bonn had to be cancelled! Read more ...

Parts | Days | Selection | Search | Updates | Downloads | Help

T: Fachverband Teilchenphysik

T 47: Neural networks and systematic uncertainties

T 47.8: Talk

Wednesday, April 1, 2020, 18:15–18:30, H-HS IV

Reinforcement learning for sorting jets in top pair associated Higgs boson production — •Dennis Noll, Martin Erdmann, and Benjamin Fischer — III. Physikalisches Institut A, RWTH Aachen University

For physics analyses with identical final state objects, e.g. jets, the correct sorting of input objects often leads to a sizeable performance increase.

We present a new approach in which a sorting network is placed in front of a classification network. The sorting network provides a two-dimensional likelihood that is used to guide the rearrangement of particle four-momenta.

Because the optimal order is generally not known, a reinforcement learning approach is chosen, in which the sorting network is trained with end-to-end feedback from the analysis. In this way, we enable the system to autonomously find an optimal solution to the sorting problem.

Using the example of top-quark pair associated Higgs boson production, we show an improvement of the signal and background separation in comparison to conventional sorting of jets with respect to their transverse momenta.