Meta-Reasoning in Adaptive Intelligent Systems Architectures
Meta-Reasoning in Adaptive Intelligent Systems Architectures
An AIS has meta-level knowledge in
its global database, and this
knowledge allows its meta-reasoning. There is information about which
operations to place on the agenda,
as well as which operations to ultimately schedule. In addition, the operations
available include those which modify perceptual and performance
filters. But perhaps the most important type of meta-reasoning
involves operators that augment and modify the dynamic control plans.
The ability to freely incorporate
meta-knowledge makes
it possible for AIS to achieve a measure of
self-reflection even though its representation of knowledge is
impenetrable. Additionally,
traces of reasoning are kept for generating
explanations, also a
form of self-reflection.
Return to the top of this architecture.
Go to a discussion of this capability
for multiple architectures.
Current Location: Adaptive Intelligent Systems-Capabilities-Meta-Reasoning