Neural network learning time: effects of network and training set size

The learning time for two-layer backpropagation networks is examined in the context of learning Boolean logic equations from examples. In particular, the relationship between the number of inputs, hidden units, and training set vectors and the learning time is investigated. The networks, the training algorithm, and the tasks are described. The parameter variations and the set of simulations performed are detailed. Training and test set generation are discussed, and the simulation results are summarized. Network performance is evaluated, and an alternate training methodology that may remedy problems inherent to the backpropagation training method is presented.<>