TITLE: Spiking Neurons with Boltzmann-like Properties to Learn Xor
SPEAKER: Christian R. Huyck (Middlesex University)

ABSTRACT:

A computational model of biological neurons is used to learn
the exclusive or relation. The neural model is a fatiguing leaky integrate
and fire model, and spontaneous firing emerges from the hypo-fatigue.
This spontaneous firing enables the system to move from direct stimulation
from the environment. A form of Hebbian learning, post-synaptic
weighted compensatory learning, is used to support firing in interior neurons.
A range of numbers of interior neurons performs the task at above
99% precision.