Theo

Reference

Mitchell, T.M., Allen, J., Chalasani, P., Cheng, J., Etzioni, O., Ringuette, M. & Schlimmer, J.C., "Theo: A framework for self-improving systems," in K. VanLehn (ed.), Architectures for Intelligence, pp. 325-355, Lawrence-Erlbaum Associates, Hillsdale, N.J., 1991.

Architecture Background

Architecture

Frames store all knowledge about beliefs and problems
Inferencing capabilities
Truth Maintenance System (TMS) maintains knowledge base
Indexing of knowledge by slot names
Three learning methods:

Agent Properties

Self-modifying
Stores meta-knowledge
Knowledge is internally consistent
Uniform representation of knowledge
Fast access to knowledge
Impasse-driven inferencing
Eager/lazy sensing
Global representation of knowledge
Can react quickly to sensor input
LARGE memory
Reflexive learning
Negative interaction among learning mechanisms
Can't non-monotonically change knowledge
All knowledge inferred from local sensing
Glass-box

Capabilities

Self-reflection
Learning
Planning/prediction
Focuses on key problems

Environment

Thinking consumes limited resources
Dynamic
Imperfect knowledge
"Big" world with lots of information
Real world

Click here to see a list of other architectures