Work Package 2

Deep Learning and HPC

‌High Performance Computing (HPC) and Deep/machine Learning is forming the research focus of this work package. The recent development of optimisation techniques that can exploit deep structures and race-tuned implementation of Deep Learning on GPUs have resulted in pervasive and successful application of Deep Learning across the Big Data arena LBDN researchers are at the forefront of this research. Numerous notable successes have changed the conventional wisdom: neural networks (and machine learning more generally) now outperform hand-crafted algorithms across a diverse range of application domains.

The following projects are part of this work package:

• Development of topological data analysis methods for AGATA 
• Deep Learning on the LHCb detector at CERN 
• LHCb Trigger based B physics analysis 
• Development of Enhanced Models of Plasma-Beam Interaction
• Self-consistent annihilation simulations of dark matter 
• Using cosmological simulations to develop large-scale structure emulators to constrain dark sector physics
• Pulse shape analysis algorithms for decay spectroscopy of short-lived nuclei
• Advanced optics concepts for HLLHC
• Betatron Radiation from Underdense Plasma
• Dielectric laser acceleration of relativistic beams
• Deep learning for LHCb with FBK
• The accretion history of the Milky Way Halo from Massive Spectroscopic Surveys and Cosmological Simulations
• Particle Acceleration in Carbon Nano Tubes
• Implementation of Machine Learning to the Muon g-2 Tracking Algorithms
• Beam Induced Fluorescence monitor for high-intensity beams