Application of a Kohonen network classifier in TeV γ-ray astronomy

A Kohonen type unsupervised artificial neural network has been used to increase the sensitivity of the atmospheric Cherenkov imaging technique used in ground-based TeV γ-ray astronomy. The network classifies Cherenkov events as γ-induced or hadron-induced on the basis of their spatial frequency components. When used in conjunction with the established Supercuts classifier it increases the sensitivity of the technique by ∼ 14%.