AI is the art of developing
and the science of understanding
computer systems that
reason,
learn,
perceive,
understand,
solve problems,
communicate,
and possess
knowledge.
Adjudipro: Health Insurance
Claim processor
Tube (pipe) inspector
Email-address finder
MYCIN infectious disease therapy
planner
PROSPECTOR expert system for minerals
exploration planning
XCON computer configuration system
for DEC VAX's
Other domains:
Machine Translation,
Handwriting Recognition,
Fingerprint Recognition,
Speech Understanding,
Mathematics Theorem Proving
Theory Development
Computer-Assisted Instruction
Search Engines
Market Research with Data
Mining
Use of knowledge
Ability to reason
Ability to learn
Ability to communicate
Ability to perceive and
understand
Intrinsic aspects:
what's going on inside
Extrinsic aspects:
qualities of the oberved behavior
What does it mean to learn?
--Acquire facts, skills, knowledge.
--Improve mental function
in some way.
--Improve behavior in some
way.
A bottle of wine put away in a wine
cellar
may improve its taste over a period
of
several years.
--Does it grow or mature?
--Does it "learn" in any
sense of the word?
The improvement should come about
as a result of information processing
activity.
When a computer downloads a file
of
information from a web site, does
it
in any way "learn" the information
in
that file?
Learning seems to involve modification
of
internal representations of knowledge
and
skills.
Learning is a relative term.
We have to
state how we are measuring the
merit of
a system in order to measure any
improvement and thereby establish
that learning may have occurred.
Function of merit: some function
that
returns a value reflecting the
ability or
quality of an information processing
system.
Inference:
Deriving new statements,
rules, and descriptions
from old.
Deductive inference:
Obtaining the new representations
by applying
logically valid rules of inference
to the old
representations.
Inductive inference:
Obtaining new general rules
and/or patterns
from specific instances or statements.
1. symbol manipulation and knowledge
representation
2. inference
3. communication, including learning
and perception.
4. design methodologies
The text includes a metaphorical
periodic table of
the elements of artificial intelligence.
The periodic
structure follows this classification
scheme.
Last modified: September 28 1998
Steve Tanimoto