Improved probabilistic neural network algorithm for chemical sensor array pattern recognition

An improved probabilistic neural network (IPNN) algorithm for use in chemic al sensor array pattern recognition applications is described. The IPNN is based on a modified probabilistic neural network (PNN) with three innovatio ns designed to reduce the computational and memory requirements, to speed t raining, and to decrease the false alarm rate. The utility of this new appr oach is illustrated with the use of four data sets extracted from simulated and laboratory-collected surface acoustic wave sensor array data. A compet itive learning strategy, based on a learning vector quantization neural net work, is shown to reduce the storage and computation requirements, The IPNN hidden layer requires only a fraction of the storage space of a convention al PNN. A simple distance-based calculation is reported to approximate the optimal kernel width of a PNN. This calculation is found to decrease the tr aining time and requires no user input. A general procedure for selecting t he optimal rejection threshold for a PNN-based algorithm using Monte Carlo simulations is also demonstrated. This outlier rejection strategy is implem ented for an IPNN classifier and found to reject ambiguous patterns, thereb y decreasing the potential for false alarms.