Abstract: Provided is a method for securing a security device against side-channel analysis attacks while performing a sensitive operation. It includes training an attack neural network to perform a side-channel attack against the security device while performing a sensitive operation, creating a training data set for a protective neural network by applying a plurality of elementary protection combinations to the sensitive operation while performing the sensitive operation, training a protective neural network executing on a coprocessor of the security device using the training data set for the protective neural network, and programming the coprocessor of the security device with the set of parameters for the protective neural network. Other embodiments disclosed.

Abstract: A neural network is trained to match digital samples to categories in a set of categories and when presented with at least one golden sample, which is a sample outside the set of categories, to output a probability vector indicative of a preposterous result that the golden sample is matched to a predefined category in the set of categories. The secure computer system is programmed with the trained neural network, adapted to receive digital samples and to present the digital samples to the trained neural network. As an integrity check, the computer system, is caused to present the golden sample to the trained neural network and if the neural network outputs a probability vector classifying the golden sample into a predefined category in a way that is a preposterous result, declaring the neural network as uncompromised and, otherwise, declaring the neural network as compromised.