KBAI 笔记 01 Intro to Knowledge-ba
Conundrums of AI
Five Conundrums of AI
- how AI agents give near real-time solutions with limited computation resources
- How local AI agent explore and address global constraints with local knowledge
- How deductive AIs solve abductive or inductive problems
- How can AI solve new problems with limited knowledge
- How we get AI to explain justifications
Characteristics of AI Problems
AI Problems
How do we design AI agents to address the characteristics?
Properties of AI agent
Properties of AI agent
Exercise: What are AI Problems?
quiz: AI problems
Exercise: AI in Practice: Watson
Quiz: what Watson can do
What is Knowledge-Based AI?
deliberation
- Reasoning, learning, and memory are closely interactive with each other and form a system called "deliberation", with is part of the cognitive system.
Cognitive system
Thinking-acting spectrum and optimal-like human spectrum
Quadrants of AI
Computation technique examples fall into the quadrants
Exercise: What is KBAI?
Where does self-driving automobile belong?
Exercise: The Four Schools of AI
What are Cognitive Systems?
Definition of Cognitive Systems
- The definition indicates that cognitive systems are in the quadrants on the left side of the spectrum.
Single cognitive systems
Percepts of the worlds go into Cognitive Systems as inputs via sensors and Cognitive systems give output with (???).
Multiple Cognitive Systems
- Cognitive Systems interact with the world and other Cognitive Systems.
Cognitive system
three-layer architecture.
- reactive system.
- deliberation
- Metacognition
Topics in KBAI
Topics overview
Fundenmentals
Planning
Common Sence Reasioning
Learning
Analogical Reasoning
Visuospatial Reasoning
Design & Creativity
Metacognition
recap
Recap
Final quiz: What have I learned from this class.
final quiz
In this course, I learned the characteristics of AI agent, AI problems, and major conundrums of AI. All the AI agents have limited computational resources, sensors attention, or knowledge and the logic they use is deductive. While the problems that AI needs to solve
are dynamic, open-ended and often computationally intractable.
There are four schools of AI based the optimally-human-like spectrum and thinking-acting spectrum. Cognitive systems are more human-like.
The cognitive system have three modules: creative, deliberation, and metacognition. and in deliberation, there are three modules: reasoning, Learning, and Memory.
