This thesis introduces a novel method for explaining the predictions of ensembles of artificial neural networks on a case by case basis. Current research is primarily directed towards building global model, that is, models ...