Berlin 2005 – scientific programme

Parts | Days | Selection | Search | Downloads | Help

DY: Dynamik und Statistische Physik

DY 22: Networks

DY 22.2: Talk

Saturday, March 5, 2005, 11:45–12:00, TU H2032

Neural Networks for Gamma-Hadron-Separation — •Christoph Kolodziejski and Georg Reents — Institiut für Theoretische Physik und Astrophysik, Universität Würzburg, Am Hubland, D-97074 Würzburg

The MAGIC-Telescope, an atmospheric imaging Cherenkov telescope in La Palma, gathers information in form of patterns of different types of particles. The interest is focused on the detection of gamma-rays. Compared to cosmic Cherenkov radiation signals they only constitute minor contributions, but look very alike. With the help of neural networks, a better separation than the currently used statistical algorithms is likely to be found. Different types of networks (e.g. Feed-Forward Multilayer Perceptrons, Pulsed Coupled Neural Nets) are analyzed to achieve this task.