Meta-Reasoning in Prodigy
Meta-Reasoning in Prodigy
Prodigy, according to its designers, decides what to learn, when to
learn, and why it learned some particular piece of knowledge. This
implies a meta-reasoning capabilty in Prodigy. This capabilty is
chiefly limited to learning control rules and how this learning
relates to a rule's utility. The STATIC
learning module accesses the contents of the agent's knowledge base to
form new control rules. Additionally, the EBL
and ANALOGY learning modules use
problem-solving traces to create new control rules based on the
agent's previous actions.
Return to the top of this architecture.
Go to a discussion of this capability
for multiple architectures.
Current Location: Prodigy-Capabilities-Meta-Reasoning