Abstract
Artificial neural networks have been employed in diverse applications ranging from control, to pattern recognition and classification. While password detection can be implemented with a digital electronic circuit with non-volatile memory, this implementation is prone to hacking. In this paper, we present a Slayer feedforward neural network which we have designed, trained and tested for secure password detection. This network correctly detects with 100% accuracy when a password presented at its inputs matches its associated user ID which the network memorized during the training phase. The BrainMaker and NetMaker tools were used for training, simulating and testing our neural network. The paper also reports our experiments with increasing the number of hidden layers, number of neurons in the hidden layer, and noise additions on the network's detection accuracy.