I am trying to validate the outputs from Statistica ANN. While comparing the network output with excel, there is a mismatch between the results (output from ANN and excel). There are two hidden layers (Hyperbolic function) and output (logistic) in the network. By applying the weights and transformation functions in excel after preprocessing of inputs, it is giving a different output as compared to the activations from ANN layers. I want to know, how the activations are being calculated in ANN?