DESIGN PRINCIPLES FOR AN INTELLIGENT MACHINE,

摘要

This paper discussed the role of prediction as the key process underlying the function of an intelligent machine. A model of a 'neuron'was presented which exhibits properties of memory and learning. The formalism of the calculus of probability allows one to interpret the behavior of a neuron in such a way as to justify how a network of such elements can be organized so that it can learn to predict. (Author)