Learning Cost in Prodigy
Learning Cost in Prodigy
Even when
deliberative learning is used to
decide what to learn in a
knowledge-rich domain, there remains a
cost associated with whatever is learned. This is a somewhat more general
case of the
Utility Problem.
One way to address this
problem is to invoke different learning strategies for different types
of knowledge to be learned, using the method best suited for the particular
problem to reduce the cost. This is exemplified in Prodigy's design by
the use of
multiple learning modules.
These modules are designed so that the knowledge they generate is
accessible to all learning modules
and the problem solver.
Return to the top of this architecture.
Go to a discussion of this environment
for multiple architectures.
Current Location: Prodigy-Environments-Learning Cost