An alternative computational model for artificial intelligence
Abstract
One objective of artificial intelligence is to mimic the behavior of natural intelligences, and learning is one of the most inherently intelligent activities. Unfortunately, there has not been much success in emulating the low-level learning of classical conditioning and instrumental learning. A novel model of computation, called the cybernetic automaton, which possesses a number of desirable properties, is presented here. First there exists a learning theorem which states that cybernetic automaton learning is in some sense complete. Then through a series of experiments, it is shown that cybernetic automata exhibit many of the properties of classical conditioning and instrumental learning. Finally, the cybernetic automaton model can be implemented using networks of model neurons which are somewhat biologically plausible. Cybernetic automata represent a new and substantial step toward artificially emulating natural behavior.
Degree
Ph.D.
Advisors
Lee, Purdue University.
Subject Area
Computer science|Artificial intelligence
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.