Brian D. Davison
While numerous node architectures have been proposed, most in use are based in part on the McCulloch-Pitts model of the neuron. One noteworthy alternative, however, is the Probabilistic Logic Neuron, advocated by Igor Aleksander and others. This thesis provides a short background in neural networks, compares the McCulloch-Pitts neuron with the Probabilistic Logic Neuron, and provides experimental results of networks using each type of neuron to solve some benchmark problems.
Published as an undergraduate honors thesis in Computer Science and Engineering, Bucknell University, 1991.
Back to Brian Davison's publications