CMPT420_01 - Web Access for Home

addictedswimmingAI and Robotics

Oct 24, 2013 (3 years and 7 months ago)

82 views

CMPT 420

Artificial Intelligence


Instructor: Tina
Tian



About me


Email:
tina.tian@manhattan.edu


Office:
RLC 203A


Office Hour: Mon,
Thur

2:00


3:30PM






or by appointment


Website
:
home.manhattan.edu/~
tina.tian


About the Course


Mon, Wed,
Thur

4:30

5:20PM


Mon,
Thur

4:30

5:20PM


Wed 4:45
-
5:35PM (15 minute delay)


Prerequisite: CMPT102 (C++)


Textbook:


Artificial
Intelligence: A Modern
Approach
(
AIMA
), 3
rd

Edition
,
by Stuart
Russell and Peter
Norvig
,
Prentice
Hall, 2010. ISBN: 0136042597


What you care the most


Grading:


1
st

Midterm Exam (in class, 5
th

week)

15%


2
nd

Midterm
Exam (in
class,
10
th

week)

15%


Final Exam





30%


Homework





20%


Project






20%


Homework


Written problems


Hard copy


Hand written homework is not accepted


Except graphs, trees, etc.


Strict deadline!


Due in a week after being announced


Late homework will have points deducted



(mailbox in RLC 203)


Not accepted
a week after the due date

Projects


Programming


C++ or Java


Submitted to Moodle (
lms.manhattan.edu
)


No partial credit
is given to projects.


Due before the last day of class (Dec 6)


What to submit


Make a nice cover page (
Readme.doc
)


Description of the problem


Algorithm chosen and why


Explanation of functions and data structures used


I
nput and output format (give an example)


Source code and executable files (
.zip
)


.
cpp

and .exe


.java and .class


zip the whole project folder
if you are using
Eclipse,
NetBeans

or VS


About Teamwork


You may
discuss
the homework/projects with
other students.


However, you must
acknowledge
the people
you worked with.


And you must
independently
write up your
own
solutions.


Any written sources used (apart from the text)
must also be acknowledged.

Advices


Keep monitors off!


Take notes


Start the homework and projects early

What AI covers


1997 game between the chess champion
Garry Kasparov and DEEP BLUE


Asimo

humanoid robot


Thomas Bayes (1702


1761)


Mars Exploration Rover (2004
-

)


Alan Turing (1912


1954)


Shakey

(1966


1972) with its project
leader Charles Rosen (1917


2002)


Aristotle (384 B.C.


322 B.C.) and his
planning algorithm in original Greek


Bayesian network for medical diagnosis

Subfields of AI


Heuristic Search


Adversarial
Search (Games)


Fuzzy Logic


Natural Language Processing


Knowledge
Representation


Computer Vision


Robotics


Planning


Learning


...

What you will learn


Introduction of AI and intelligent agents


Searching algorithms (Chap 3
-
6)


Uninformed search, informed search, local search, game
tree, backtracking ..


Rule
-
based expert system


forward and backward
chaning


Uncertainty in AI



Bayes’ rule, Bayesian networks, certainty factor, fuzzy logic


Machine learning


Decision tree, neural networks


Knowledge representation and data mining


Ontological engineering, association rule


Reading


AIMA Chapter 1

What is Artificial
Intelligence
?


Essential English Dictionary
, Collins, London, 1990:


Someone’s
intelligence

is their ability to understand
and learn things.


Intelligence

is the ability to think and understand
instead of doing things by instinct or automatically.


Thinking
is the activity of using your brain to
consider a problem or to create an idea.


We can define
intelligence
as ‘the ability to learn
and understand, to solve problems and to make
decisions’.


What is Artificial Intelligence?


Psychological approach: an intelligent system
is a model of human intelligence



Engineering approach: an intelligent system
solves a sufficiently difficult problem in a
generalizable way


What is Artificial Intelligence?

(
again)


Systems that think like
humans


Cognitive Modeling Approach


“The automation of activities
that we associate with human
thinking...”


Bellman 1978



Systems that act like
humans


Turing Test Approach


“The art of creating machines
that perform functions that
require intelligence when
performed by people”


Kurzweil 1990


Systems that think
rationally


“Laws of Thought” approach


“The study of mental faculties
through the use of
computational models”


Charniak

and McDermott



Systems that act rationally


Rational Agent Approach


“The branch of CS that is
concerned with the
automation of intelligent
behavior”


Lugar and Stubblefield


Thinking Humanly


1960’s cognitive revolution


Requires scientific theories of internal activities of
the brain


What level of abstraction? “Knowledge” or “Circuits”


How to validate?


Predicting and testing behavior of human subjects (top
-
down)


Direct identification from neurological data (bottom
-
up)


Cognitive Science and Cognitive Neuroscience


Now distinct from AI

Acting Humanly: The Turing Test


Alan Turing, British mathematician (1912
-
1954)



Computing machinery and intelligence
” paper in 1950


Can machines think?


The
Turing Test

(a.k.a.
Turing imitation game
):





Predicted
that by 2000, a machine might have a 30% chance
of fooling a lay person for 5 minutes




A computer passes the Turing test if human
interrogators cannot distinguish the
machine from a human based on answers to
their
questions

The Turing Test


Alan
Turing suggested an imitation
game.


A person C questions two other
“agents” A and B over a computer
terminal.


The person C cannot see or hear A
and B.


Both A and B claim they are
humans.


But one of them is lying.


If C cannot detect that A is a
computer, that means that A is for
all practical purposes “intelligent.”

Human

AI System

Human

Interrogator

?

Loebner

Prize


The
Loebner

Prize is an annual competition
for AI programs.


http://
www.loebner.net/Prizef/loebner
-
prize.html


Crown Industries of East Orange, NJ


$100,000 and a Gold Medal for the first
computer that passes the Turing Test.


Each year $2000 and a bronze medal is
awarded to the most human
-
like computer.

The Turing Test


Natural language processing


Knowledge representation


Automatic reasoning


Machine learning


Total Turing Test: computer vision and
robotics

Thinking Rationally


Aristotle
: What are correct arguments / thought
processes
?


“Socrates is a man; all men are mortal; therefore, Socrates
is mortal.”


Logic notation and rules for derivation for thoughts


Problems:


Not all
knowledge can be transformed to logic, especially
when it is less than 100% certain


Problems with a few hundred facts can exhaust
computational resources


Direct line through mathematics and philosophy to
modern AI

Acting Rationally


Rational

behavior


Doing the right thing


What
is the “right thing”


That which is expected to maximize goal achievement,
given available information


We
do many (“right”) things without thinking


e.g., blinking reflex


Thinking
should be in the service of rational
action


The textbook advocates "acting rationally"

Rational agents


An
agent
is an entity that perceives and
acts


This course is about designing rational
agents


Abstractly, an agent is a function from percept
histories to actions
:

[
f
:
P*



A
]


For any given class of environments and tasks, we
seek the agent (or class of agents) with the best
performance


Caveat: computational limitations make perfect
rationality
unachievable


design
best

program

for given machine
resources

Strong AI vs. Weak AI


Strong AI
is artificial intelligence that matches
or exceeds human
intelligence.


“Artificial
general
intelligence”


The
weak AI
hypothesis:
machines
can
demonstrate intelligence, but do not
necessarily have a mind, mental states or
consciousness.

Chinese Room Argument


John Searle


http://
en.wikipedia.org/wiki/John_Searle


“Chinese room” setup






Searle answers questions given to him in Chinese
(though he does not know any Chinese)

Searle

+

book of rules

Chinese Question

Answer in Chinese

Chinese Room Argument





Book
: instructions to answer Chinese questions


Searle
: just applies algorithm given in book


Does Searle now "understand" Chinese?


Observation: every computer can be described on
paper


Searle

+

book of rules


恭喜发财”

“谢谢”

Chinese Room Argument


Strong AI advocates: it's Searle + room
+
book
that understands!


reply
: suppose Searle memorizes algorithm
in
his
head: does he understand now?


executing an algorithm does

not

constitute thinking


the terms "understand" and "think" are not
something we can apply to inanimate
objects


"
Minds, Brains, and Programs
" by John R. Searle
(
The Behavioral and Brain Sciences
,
Vol

3 (1980))



AI prehistory


Philosophy


Logic, methods of reasoning, mind as physical





system foundations of learning, language,




rationality


Mathematics


Formal representation and proof algorithms,




computation, probability


Economics


utility, decision theory
(to maximize payoff)


Neuroscience


physical substrate
for mental activity


Psychology


phenomena of perception and motor control,




How do humans and animals think and act?


Computer


building fast computers

engineering


Linguistics


knowledge representation, grammar

History of AI


Warren McCulloch & Walter Pitts (1943):


Research on the human central nervous system
led to a model of neurons of the brain


Birth of
Artificial Neural Networks

(
ANN
)


Binary model


Non
-
linear model



John von Neumann


ENIAC, EDVAC, etc.

History of AI


Claude Shannon, MIT, Bell Labs (1950):


Computers playing chess


Chess game involved about 10
120

possible moves!


Even examining one move per
microsecond

would
require 3
x

10
106

years to make its first move



Need to incorporate intelligence via
heuristics

History of AI


John McCarthy, Dartmouth, MIT (1950s):


Defined LISP


Only two years after FORTRAN


LISP is based on formal logic



Programs with Common Sense
” paper (1958)



Marvin
Minsky
, Princeton, MIT:


Anti
-
logical approach to knowledge
representation and reasoning called
frames

(1975)

History of AI


Great expectations during 1950s and 1960s


But very limited success


Researchers focused too much on all
-
purpose
intelligent machines with goals to learn and
reason with human
-
scale knowledge (and
beyond)



Refocus on specific problem domains (1970s)


Domain
-
specific
expert systems

with
facts
,
rules
, etc.


Analyze chemicals, medical diagnoses, etc.

History of AI


Evolutionary computation (1970s
-
today):


Natural intelligence is a product of evolution


Can we solve problems by simulating

biological evolution?


Survival of the fittest


Genetic programming


Evolutionary computing

History of AI


Rebirth of neural networks (1980s
-
today):


Adaptive resonance theory

(
Grossberg
, 1980)
incorporated self
-
organization principles


Hopfield networks

(Hopfield, 1982)
introduced neural networks with
feedback loops


Back
-
propagation learning algorithm

(Bryson and Ho, 1969) for training
multilayer
perceptrons


History of AI


Knowledge engineering (1980s
-
today):


Fuzzy set theory

(
Zadeh
, 1965) associates words

with degrees of truth or value


Rule
-
based knowledge systems


Combine information from multiple experts


Semantic Web



Numerous hybrid approaches exist

Abridged
History
of AI


1943

McCulloch & Pitts: Boolean circuit model of brain


1950

Turing's "Computing Machinery and Intelligence"


1956


Dartmouth
meeting: "Artificial Intelligence" adopted


1952

69

Look, Ma, no hands!


1950s

Early AI programs, including Samuel's checkers



program, Newell & Simon's Logic Theorist,



Gelernter's Geometry Engine


1965


Robinson's
complete algorithm for logical reasoning


1966

73

AI discovers computational complexity



Neural network research almost disappears


1969

79

Early development of knowledge
-
based systems


1980
--


AI becomes an industry


1986
--


Neural networks return to popularity


1987
--

AI becomes a science


1995
--

The emergence of intelligent agents

State of the
Art


Game playing
: IBM’s Deep
Blue defeated the reigning world
chess champion
Garry Kasparov in
1997.


Speech recognition
: A traveler calling United Airlines to book
a flight can have the entire conversation guided by an
automatic speech recognition system.


Robotic vehicles
: No
hands across America (driving
autonomously 98% of the time from Pittsburgh to San Diego)


Logistics planning
: During
the 1991 Gulf War, US forces
deployed an AI logistics planning and scheduling program that
involved up to 50,000 vehicles, cargo, and
people.


Autonomous planning and scheduling
: NASA's
on
-
board
autonomous planning program controlled the scheduling of
operations for a
spacecraft.


Robotics
: The iRobot Corporation has sold over two million
Roomba robotic vacuum cleaners for home use.

Less Successful Areas of AI


Sadly the
Loebner

Gold Medal still has not
been awarded.


Natural Language Processing is still mostly an
unresolved problem.


Ninety/Ten
Rule
: Can
do 90% of the translation
with 10% time, but 10% work takes 90% time


Can this be solved by computers?


Playing a decent game of table tennis (Ping
-
Pong).


Can this be solved by computers?


Playing a decent game of table tennis (Ping
-
Pong).


A reasonable level of proficiency was achieved by
Andersson’s

robot (
Andersson,1988
).

Can this be solved by computers?


Driving in the center of Cairo, Egypt.


Can this be solved by computers?


Driving in the center of Cairo, Egypt.


No. Although there has been a lot of progress in
automated
driving, all
such systems currently rely on
certain relatively constant clues: that the road
has
shoulders
and a center line, that the car ahead will travel
a predictable course, that
cars will
keep to their side of
the road, and so on. Some lane changes and turns can be
made on
clearly marked roads in light to moderate
traffic. Driving in downtown Cairo is
too unpredictable
for any of these to work.

Can this be solved by computers?


Buying a week’s worth of groceries at the
market.


Can this be solved by computers?


Buying a week’s worth of groceries at the
market.


No. No robot can currently put together the tasks of moving
in
a
crowded environment, using vision to identify a wide variety
of objects, and
grasping the
objects (including
squishable

vegetables) without damaging them. The
component pieces
are nearly able to handle the individual tasks, but it would
take a major
integration effort
to put it all together.

Can this be solved by computers?


Buying a week’s worth of groceries on the
Web.


Can this be solved by computers?


Buying a week’s worth of groceries on the
Web.


Yes. Software robots are capable of handling such tasks,
particularly if
the design of the web grocery shopping site
does not change radically
over time
.

Can this be solved by computers?


Writing an intentionally funny story.


Can this be solved by computers?


Writing an intentionally funny story.


No. While some computer
-
generated prose and poetry is
hysterically funny
, this is invariably unintentional, except in
the case of programs that echo
back prose
that they have
memorized.

Unintentionally
F
unny Stories


One day Joe Bear was hungry. He asked his
friend Irving Bird where
some honey
was.
Irving told him there was a beehive in the oak
tree. Joe
threatened
to hit Irving if he didn't
tell him where some honey was. The End.


Henry
Squirrel was thirsty. He walked over to
the river bank where his
good friend
Bill Bird
was sitting. Henry slipped and fell in the river.
Gravity drowned
. The End.

Can this be solved by computers?


Giving competent legal advice in a specialized
area of law.


Can this be solved by computers?


Giving competent legal advice in a
specialized area of law.


Yes, in some cases. AI has a long history of research into
applications of
automated legal reasoning.
One example
is the
Prolog
-
based
expert systems
used in the UK to
guide members of the public in dealing with the
intricacies
of the
social security and nationality laws.
However, extension
into more complex areas such as
contract law awaits a satisfactory
encoding of
the vast
web of common
-
sense knowledge pertaining to
commercial transactions
and agreement
and business
practices
.

Can this be solved by computers?


Translating spoken English into spoken
Swedish in real time
.

Can this be solved by computers?


Translating spoken English into spoken
Swedish in real time.


Yes. In a limited way, this is already being done.

Can this be solved by computers?


Performing a complex surgical operation.


Can this be solved by computers?


Performing a complex surgical operation.


Yes. Robots are increasingly being used for surgery, although
always
under the
command of a doctor. Robotic skills
demonstrated at superhuman levels
include drilling
holes in
bone to insert artificial joints, suturing, and knot
-
tying. They
are
not yet
capable of planning and carrying out a complex
operation autonomously from
start to
finish.

Reading


AIMA Chap 2


"
Computing Machinery and Intelligence" by
Alan Turing (
Mind
, Vol. LIX, No. 236 (1950)).


"Minds, Brains, and Programs" by John R.
Searle (
The Behavioral and Brain Sciences
,
Vol

3 (1980
))