An Introduction to Artificial Intelligence

trainerhungarianAI and Robotics

Oct 20, 2013 (4 years and 19 days ago)

74 views

An Introduction to Artificial
Intelligence

Introduction


Getting machines to “think” .


Imitation game and the Turing test.


Chinese room test.


Key processes of AI:


Search, e.g. breadth first search, depth first
search, heuristic searches.


Knowledge representation, e.g. predicate
logic, rule
-
based systems, semantic
networks.

Areas of AI


Game playing


Theorem proving


Expert systems


Natural language processing


Modeling human performance


Planning and Robotics


Neural
-
networks


Evolutionary algorithms and other
biologically inspired methods


Agent
-
based technology


Game Playing


Getting the computer to play certain
board games that require “intelligence”,
e.g. chess, checkers, 15
-
puzzle.


A state space of the game is developed
and a search applied to the space to
look ahead.


Example: Deep blue vs. Kasparov.

.

Theory Proving


Automatic theorem proving.


Generate proofs for simple
theorems.


Mathematical logic forms the basis
of these systems.


The “General Problem Solver” is one
of the first systems.


.

Expert Systems


Performs the task of a human
expert, e.g. a doctor, a psychologist.


Knowledge from an expert is stored
in a knowledge base.


Examples: ELIZA, MYCIN, EMYCIN


Suitable for specialized fields with a
clearly defined domain.

.

Natural Language Processing


Develop systems that are able to
“understand” a natural language
such as English.


Voice input systems, e.g. Dragon.


Systems that “converse” in a
particular language.


Examples: SHRDLU and ELIZA

.

Modeling Human Performance


Systems that model some aspect
of problem solving.


Examples: Intelligent tutoring
systems that provide
individualized instruction in a
specific domain.



.

Planning and Robotics


Involves designing flexible and
responsive robots.


Lists of actions to be performed
are generated.


Aimed at high
-
level tasks, e.g.
moving a box across the room.


Has led to agent
-
oriented
problem solving.

Neural Networks


Aimed of low
-
level processing.


Are essentially mathematical models of
the human brain.


A neuron:

.

Evolutionary Algorithms & Other
Nature
-
Inspired Algorithms



Based on Darwin’s theory of evolution.


An initial population of randomly created
individuals is iteratively refined until a
solution is found.


Examples: genetic algorithms, genetic
programming,
memetic

algorithms


Other methodologies: ant colonization,
swarm intelligence.

.

Uncertainty Reasoning


Uncertain terms may need to be
presented.


Example: representing terms such as
“big” or “small”.


Methods for this purpose:


Fuzzy logic


Bayesian reasoning and networks

.

Agent
-
based Technology


Intelligent agents, also called

softbots
”, are used to perform
mundane tasks or solve
problems.


In a multi
-
agent system agents
communicate using an agent
communication language.

.

Artificial Intelligence Languages


Programming paradigms


Artificial intelligence languages


Prolog
and Lisp


Prolog (
Pro
gramming
Log
ic)


declarative


predicate logic


Lisp (
Lis
t
P
rocessing)


functional


code
takes the form of recursive functions.


More recently AI systems have been
developed in a number of languages
including Smalltalk, C, C++ and Java.