Application of an artificial neural network in the SALC explosive detection system

Summary form only given. The detection of explosives in checked airline luggage by thermal neutron analysis has previously been shown to be improved by the use of an artificial neural network in place of classical decision techniques. These previous studies have focused on the end result (overall detection and false alarm rates); in the work presented, a small portion of the data was examined to show the decision surfaces of the two techniques. The relative misclassifications of the two techniques were compared.<
>