Accelerated machine learning for collider and neutrino physics

In this project, Aleksander will develop machine learning-based algorithms for data compression, signal classification and outlier detection to use in data produced by the Large Hadron Collider and neutrino experiments, and to prepare for upcoming collider and neutrino experiments. The main tool used in this project will be the open-source software "Baler" available at https://github.com/baler-collaboration/baler.

This tool uses machine learning to derive a compression method that is tailored to the user's input data, achieving large data reduction with a minimal fidelity loss.

In this project, Aleksander will be applying this compression tool and benchmark its performance on open data made available from the collider and neutrino experiments. The main goals of this project will be to bring this algorithm to production for the upcoming data taking periods of the experiments, and includes improving the performance of the algorithm itself, applying it to real-time analysis (triggers), implementing it on custom electronic boards (FPGAs) and demonstrating its performance on physics analysis.