Neural nets for ground-based γ-ray astronomy

The application of the artificial neural networks to data processing in the domain of the atmospheric Cherenkov γ-ray telescopes is considered. The main problems arising from the specifics of these experiments, such as low signal, pairs of ON–OFF observations, instability of the background events, and their influence on the analysis are discussed. A method for discriminating the γ-induced atmospheric showers from the large hadronic background is proposed.