Multi-Method Learning in Meta-Reasoning Architectures
Multi-Method Learning in Meta-Reasoning Architectures
MAX has no architecturally defined learning mechanism. However,
learning is facilitated by the uniform representation of all
knowledge and reasoning behavior. Each learning mechanism implemented
as a behavior. Any number of
learning mechanisms may be incorporated; currently MAX uses:
- exploration
- experimentation
- behavior debugging
- explanation-based learning
Because of the sequential application of
behaviors, only one learning mechanism can be active at any time,
and each mechanism must be explicitly invoked, giving rise to
deliberative learning.
Return to the top of this architecture.
Go to a discussion of this capability
for multiple architectures.
Current Location: Meta-Reasoning-Capabilities-Multi-Method-Learning