Artificial Intelligence: A Modern Approach

1
Introduction

• Artificial Intelligence A Modern Approach
• (Chapter 1, 26)
• Juntae Kim
• Department of Computer Engineering
• Dongguk University

2
What is Intelligence?

• Perception
• Vision (Image understanding)
• Speech (language understanding)
• Reasoning
• Logical inference
• Given facts ? new facts
• Learning
• Memory, generalization
• Improving performance as it repeats

3
What is Artificial Intelligence?

• Study of how to make computer do things that
  people do better
• To find how people think/act intelligently
• To develop systems that perform intelligent task

4
Some Definitions
Thinking
Cognitive modeling Automation of human thinking
(decision making, problem solving, learning)
Logic Study of mental faculties through the use
of computational model
Human-like
Rational
Turing test approach Make computers do things at
which people are better (NLP, reasoning, vision,)
Rational agent The branch of CS that is concerned
with the automation of intelligent behavior
Acting
5
Acting Humanly

• The Turing test approach
• Interrogated by a human via teletype
• ? passes if the interrogator cannot tell if
  there is a computer or a human
• Necessary capabilities
• Natural language processing
• Reasoning
• Learning
• Vision, robotics
• Internal representation/processing may not be
  based on human model

6
Thinking Humanly

• Cognitive modeling approach
• Make programs that think like human
• ? need to study how human mind works
• GPS(General Problem Solver)
• More concerned with the programs reasoning steps
  comparing it to that of humans
• Cognitive science
• Construct theories of the workings of human mind
• Computer models from AI experimental techniques
  from psychology

7
Thinking Rationally

• The law of thought approach
• Study the right thinking(reasoning) process
• ? logic and inference mechanism
• Reasoning system based on logic
• Well defined, well understood
• Difficulties
• Informal knowledge ? logical notation

8
Acting Rationally

• The rational agent approach
• Building a rational agent
• Agent something that perceive and act
• Rational act so as to achieve ones goals
• Logical inference, language processing, vision,
  learning can be a part of rational agent
• Some rational action may not involve inference
• Exgt Pulling ones hand off of a hot stove

9
History of AI

• Gestation
• Artificial neuron (McCulloch Pitts, MIT)
• Logic Theorist (Newell Simon, CMU)
• Dartmouth workshop ? Artificial Intelligence
• Early enthusiasm
• Checker program (Samuel, IBM)
• GPS (Newell Simon, CMU)
• ELIZA (Weizenbaum, MIT)
• Machine Translation Project (NRC)

10
History of AI

• Knowledge-based systems
• DENDRAL (Buchanan, Stanford)
• MYCIN (Feigenbaum, Stanford)
• Commercial expert systems
• 5th generation computer project (Japan)
• Current state of the art
• Practical applications
• Statistical approaches

11
Areas/Applications of AI

• Reasoning
• VLSI verification, Fuzzy control
• Expert Systems
• Medical, financial, design, scheduling, law
• Vision
• Face recognition, autonomous vehicle
• Speech Recognition
• Voice recognition, speech translation
• Natural Language Understanding
• Information extraction, machine translation

12
Areas/Applications of AI

• Machine Learning
• Neural networks, data mining, classification,
  prediction
• Robotics
• Motion planning, assembly planning
• Game Playing
• Chess, NPC control
• Web agents
• Personal desktop agent, recommender systems,
  e-commerce agent

13
Turing Test

• Alan Turing (1950) - Can machine be intelligent?
• 2004 Turing test (Loebner contest)
• http//www.alicebot.org/ ? Chat with A.L.I.C.E

14
Conversation with Robots

• KISMET
• A robot made in MIT AI lab
• Vision object recognition, movement detection
• Speech recognition
• Attention/Behavior control
• Speech generation
• Introducing emotions, belief, desire
• Expression examples
• Focusing
• Dialog with humans

15
The Chinese Room Experiment

• John Searl (1980)
• Experiment
• Searls argument
• The man in the room does not have any
  understanding
• No matter how intelligent-seeming a computer
  behaves and no matter what programming makes it
  behave that way, since the symbols it processes
  are meaningless (lack semantics) to it, it's not
  really intelligent. It's not actually thinking.

16
The Brain Prosthesis Experiment

• John Searl (1980),
• Hans Moravec (1988)
• Experiment
• Will you remain conscious?
• Hans Moravec(1988) Yes. It will show exactly
  same behavior with real brain. The mechanism
  involved in consciousness are still operating in
  the electronic brain, which is therefore
  conscious
• John Searl(1980) No. You would end up loosing
  your control over your behavior. (Problem The
  consciousness cannot be removed instantaneously)

17
The Worlds First Brain Prosthesis Artificial
Hippocampus
18
Can Machines Think?

• Am I thinking or what?
• Intelligence? (weak AI)
• yes, in narrow sense.
• Consciousness? (strong AI)
• debate unsolved.

19
EasyDiagnosis Expert System
20
EasyDiagnosis Expert System
21
MAPGEN (Mixed Initiative Activity Planning
Generator)
22
SHYSTER
23
Autonomous Vehicles

• Mars Autonomy - Long-distance marsrover
  navigation with minimal human intervention.

• UGV - unmanned ground vehicle for reconnaissance,
  surveillance, and combat.

24
ECTACOs Voice Translator for Pocket PC
25
ezTrans by ChangShin Soft
26
Classification of Tumors
27
Brain MRI Classification
28
IBM Deep Blue
29
NPC Control
30
Recommender Systems