SPASE-AMANDA coincident event analysis

Neural Net

TMultiLayerPerceptron in ROOT is used for this study. NN is trained with old SPASE-AMANDA MC.

Probability map as a function of observables (S30, K50)

True energy Vs. Reconstucted energy

The old method?

Energy resolution for proton and iron showers (old method)

Energy resolution for proton and iron showers (NN)

NN with more input variables is under way.