Tuning an artificial neural network to increase the efficiency of a fingerprint matching algorithm

In this paper we presented a simplified solution of an artificial neural network which matches the patterns of fingerprints. As universal approximator, the artificial neural networks (ANNs) are able to bridge the non linear outcome of the superposition of elementary failures. We built up a multilayer feedforward network, and used the back-propagation on a medium size sample, to reach better accuracy. This implementation needs a significantly shorter processing time and less complex IT background. We were focusing on the testing of the training's outcome, because it highly depends on the starting conditions.