Transcript The Foundations of AI and Intelligent Agents

Computer Science & Engineering, University of Nevada, Reno
CS482/682
Artificial Intelligence
Lecture 1:
The Foundations of AI
and Intelligent Agents
25 August 2009
Instructor: Kostas Bekris
482/682
What is AI?
Humanly
vs.
Rationally
Thinking
“The automation of activities that
“The study of mental faculties
we associate with human thinking,
through the use of computational
activities such as decision-making,
models.”
problem solving, learning”
(Winston, 1992)
(Bellman, 1978)
vs.
“AI is concerned with rational
“The art of creating machines that
action... and studies the design of
perform functions that require
rational agents. A rational agent
intelligence when performed by
acts so as to achieve the best
people”
expected outcome”
(Kurzweil, 1990)
(S.R. & P.N., 1995)
Acting
482/682
What is AI?
Humanly
vs.
Rationally
Thinking
“The automation of activities that
“The study of mental faculties
we associate with human thinking,
through the use of computational
activities such as decision-making,
models.”
problem solving, learning”
(Winston, 1992)
(Bellman, 1978)
vs.
“AI is concerned with rational
“The art of creating machines that
action... and studies the design of
perform functions that require
rational agents. A rational agent
intelligence when performed by
acts so as to achieve the best
people”
expected outcome”
(Kurzweil, 1990)
(S.R. & P.N., 1995)
Acting
482/682
Acting Humanly
482/682
What is AI?
Humanly
vs.
Rationally
Thinking
“The automation of activities that
“The study of mental faculties
we associate with human thinking,
through the use of computational
activities such as decision-making,
models.”
problem solving, learning”
(Winston, 1992)
(Bellman, 1978)
vs.
“AI is concerned with rational
“The art of creating machines that
action... and studies the design of
perform functions that require
rational agents. A rational agent
intelligence when performed by
acts so as to achieve the best
people”
expected outcome”
(Kurzweil, 1990)
(S.R. & P.N., 1995)
Acting
482/682
Thinking Humanly
482/682
What is AI?
Humanly
vs.
Rationally
Thinking
“The automation of activities that
“The study of mental faculties
we associate with human thinking,
through the use of computational
activities such as decision-making,
models.”
problem solving, learning”
(Winston, 1992)
(Bellman, 1978)
vs.
“AI is concerned with rational
“The art of creating machines that
action... and studies the design of
perform functions that require
rational agents. A rational agent
intelligence when performed by
acts so as to achieve the best
people”
expected outcome”
(Kurzweil, 1990)
(S.R. & P.N., 1995)
Acting
482/682
Thinking Rationally
482/682
What is AI?
Humanly
vs.
Rationally
Thinking
“The automation of activities that
“The study of mental faculties
we associate with human thinking,
through the use of computational
activities such as decision-making,
models.”
problem solving, learning”
(Winston, 1992)
(Bellman, 1978)
vs.
“AI is concerned with rational
“The art of creating machines that
action... and studies the design of
perform functions that require
rational agents. A rational agent
intelligence when performed by
acts so as to achieve the best
people”
expected outcome”
(Kurzweil, 1990)
(S.R. & P.N., 1995)
Acting
482/682
Acting Rationally
482/682
Intelligent Agents
482/682
Environments and their properties
482/682
Environments and their properties
482/682
Structure of the Course
Part 1.
Decision-Making in Deterministic Environments
• Single-agent: Dynamic programming and search, informed search and
heuristics, randomized search, genetic algorithms, constraint satisfaction
and path planning
• Multi-agent: Adversarial search (mini-max and expecti-mini-max)
Part 2.
Decision-Making in Stochastic Environments
• Single-agent: Bayesian networks, Hidden Markov Models, Kalman and
Particle filters, Decision and Utility theory, Markov Decision Processes
• Multi-agent: Introduction to Game Theory
Part 3.
Learning in Unknown Environments
• Supervised learning: Decision trees, Support Vector Machines, Neural
Networks
• Unsupervised learning: Introduction to Reinforcement Learning
482/682
482/682
Where are we now?
482/682
What are my personal interests?
Physicall
yGrounde
d
Agents
Robotics
Computer
Games
Human
Assistants
Agents that must and do appropriately model and reason about the
physical properties of their environment:
• algorithmic generation of motion (motion planning)
• state estimation problems given noisy sensors
• and distributed message-passing coordination
482/682
How do agents work?
482/682
Reflex Agents
482/682
Model-based Reflex Agents
482/682
Goal-based Agents
482/682
Utility-based Agents
482/682
Learning Agents