Theo Methodological Assumptions
Theo was designed as a framework to support "the next generation" of
machine learning algorithms. The developers of Theo believed that
a general, flexible knowledge base would allow agents to learn
facts and meta-facts. Meta-information
is useful for many purposes, such as the reliability, utility, and indexing
of information.
In Theo, learning is interleaved with
general problem solving and
self-reflection . This allows the architecture to react
appropriately when the environment presents opportunities for learning
or self-reflection.
An additional assumption required by Theo's truth justification
system is that the entire environment must be visible to Theo's sensors.
This is due to the automatic retraction of environmental features
when they pass out of sensing range.
Go to General Discussion of
Methodological Assumptions
Return to the top of this architecture.
Current Location: Theo-
Philosophy and Methodological Assumptions