...
- 07.04. Inga Strumke (AI Lab, NTNU): Introduction to machine learning - a guided tour with examples from particle physics
Abstract: Inga will give us a crash course in machine learning, how it's done and which forms exist. She will take us through a a few examples from HEPP, demonstrating current applications and their challenges. Finally, the million dollar question of how to explain machine learning models - also referred to as the "black box problem" - is disseminated and an overview of the status of explainable AI (XAI) is given. The talk is open for everybody, and intellectually available to anyone comfortable with arithmetic and the existence of the Higgs boson :-)
slides
- 12.05., Patrick Reichherzer (RUB Bochum): Influence of diffusive cosmic-ray transport on multimessenger observables
Abstract: Cosmic-ray transport in astrophysical environments is often dominated by diffusion in a magnetic field with a turbulent component. The diffusion properties of charged particles directly influence observable properties, such as the spectrum of cosmic rays and their secondaries produced in interactions. In many diffusion scenarios, the simplified assumption of fully resonant Kolmogorov diffusion in the quasi-linear limit results in a parallel diffusion coefficient D ~ E^(1/3). A quantitative investigation of the scattering regimes, however, shows that the diffusion coefficient tensor can deviate significantly from this behaviour. In this talk, the complex dependencies of charged particle diffusion on the turbulence level of the magnetic field are presented. Examples of how this affects observational signatures will be shown in the context of galaxies or the transient sky, i.e., flaring Blazars.
...