STELLA: A scheme for a learning machine

Summary A scheme for a learning machine is described. In a basic exploratory mode the machine searches its environment. Learning enables it to profit from those action sequences which lead to reward. By correlating the changes in its environment with its actions, the machine can extract invariant features and use them to guess profitable actions when its learned sequences fail. An internal mode of operation is described in which the machine explores the possibilities of its future actions with a view to modifying its performance. The learning automaton, STELLA, is being constructed in the form of a mechanical tortoise which takes its name from its laboratory origin.

[1]  W. Pitts,et al.  What the Frog's Eye Tells the Frog's Brain , 1959, Proceedings of the IRE.