Explanation_Based_Learning
The problem being discussed is what object could substitute for a cup if no cups are available...
Concept Space
How does an agent determine if an object can work as a cup:
Prior knowledge of cups:
- Object -> has -> Bottom
- Bottom -> is -> Flat
- Object -> made-of -> Porcelain
- Object -> has -> Decoration
- Object -> has -> Concavity
- Object -> is -> Light
- Object -> has -> Handle
Agent wants to prove:
- Object -> is -> Cup
A prior knowledge relationship looks like this:
Where the yellow arrows indicate causality
Abstraction
AI agent has to abstract from other concepts to a concept that it can use for any object.
1) A cup must be stable 2) A brick is stable because it has a flat bottom 3) A cup is stable because it has a flat bottom, like a brick
