Speaker
Description
The highly energetic nature of gamma rays renders their terrestrial observation impossible, as they interact with the atmosphere, producing particle showers. These particles then become a source of Cherenkov radiation, subsequently detected by an array of ground-based telescopes. Their cameras record images which are then parametrized. The parameters are fed to random forests to reconstruct the direction, energy and type of the primary. My work explores alternative methods that involve training of convolutional neural networks that do not require image parametrization and therefore exploit all the information available. The resulting data analysis pipeline, called CTLearn, achieves better performance which allows full exploitation of the SST1M telescopes potential.
