Analysis of hidden units in a layered network trained to classify sonar targets

-A neural network learning procedure has been applied to the classification ~/sonar returns [kom two undersea targets, a metal cylinder and a similarly shaped rock. Networks with an intermediate layer ~/ hidden processing units achieved a classification accuracy as high as 100% on a training set of l04 returns. These net~orks correctly classified up to 90.4% of 104 test returns not contained in the training set. This perfi~rmance was better than that of a nearest neighbor classifier, which was 82.7%. and was close to that of an optimal Bayes classifie~ Specific signal features extracted by hidden units in a trained network were identified and related to coding schemes in the pattern of connection strengths between the input and the hidden units. Network perlbrmance and class[/~cation strategy was comparable to that of trained human listeners. Keywords--Learning algorithms, Hidden units. Multilayered neural network, Sonar, Signal processing.