Download Artificial Intelligence

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Document related concepts

Machine learning wikipedia, lookup

Computer vision wikipedia, lookup

Human–computer interaction wikipedia, lookup

Artificial intelligence in video games wikipedia, lookup

Technological singularity wikipedia, lookup

Computer Go wikipedia, lookup

Embodied cognitive science wikipedia, lookup

AI winter wikipedia, lookup

Knowledge representation and reasoning wikipedia, lookup

Intelligence explosion wikipedia, lookup

Philosophy of artificial intelligence wikipedia, lookup

Ethics of artificial intelligence wikipedia, lookup

Existential risk from artificial general intelligence wikipedia, lookup

History of artificial intelligence wikipedia, lookup

Transcript
Artificial Intelligence
ARTIFICIAL INTELLIGENCE
Venkata Phani Anne
IMSC500-1902-Information System Technology
Professor Soroushi
The University of Northern Virginia
March 18, 2010
Page 1
Artificial Intelligence
ABSTRACT
This paper describes the development of an application of Artificial Intelligence
(AI) in game playing, speech appreciation, understanding natural language and
Computer version. And describes the solution of specific problems, longstanding
differences of opinion about how AI should be done and the applications of widely
differing tools. The most problems of AI include lot of verities as Communication,
development, learning, understanding, ability to move and acuity.
Page 2
Artificial Intelligence
CONTENENTS
Sl.
No.
1.0
2.0
3.0
3.1
3.2
3.3
3.4
3.5
3.6
3.7
4.0
4.1
4.2
4.3
4.4
4.5
5.0
5.1
5.2
5.3
5.4
5.5
6.0
6.1
6.2
6.3
6.4
6.5
6.6
7.0
8.0
Description
Description
History
Problems
Learning
Motion and manipulation
Deduction
Reasoning
General intelligence
Planning
Social intelligence
Approaches
Integrating approaches
Brain simulation
cognitive architectures
Knowledge based
Symbolic
Tools
Probabilistic method
Search and optimization
Logic
Classifiers and statistical learning methods
Evaluating progress
Applications
Game playing
Speech recognition
Understanding natural language
Computer vision
Expert systems
Heuristic classification
Uses
Future
References
Page No.
4
4
5
5
6
6
6
6
7
7
7
7
8
8
8
9
9
9
10
11
12
12
13
13
14
14
14
14
15
15
16
18
Page 3
Artificial Intelligence
1.0 DEFINITION:Artificial intelligence is a field of science and engineering concerned with the
computational perceptive of what is commonly called intelligent behavior, and with the creation
of artifacts that exhibit such behavior.
2.0 HISTORY:The history of AI is a history of fantasies, prospective, demonstrations, and guarantee.
Ever since Homer wrote of mechanical “tripods” waiting on the gods at dinner, imagined
mechanical assistants have been a part of our traditions. However, only in the last half century
have we, the AI community, been able to build untried machines that test hypotheses about the
mechanisms of thought and intelligent behavior and thereby display mechanisms of thought and
intelligent behavior and there by demonstrate mechanisms that before existed only as theoretical
potential. Although achieving full-blown artificial intelligence remains in the ongoing
conversation about the implications of realizing the promise.
Philosophers have floated the opportunity of intelligent machines as a fictitious device to
help us define what it means to be human. Rene Descartes, for example, seems to have been
more interested in “mechanical man” as a metaphor than as a possibility. Gottfried Wilhelm
Leibniz, on the other hand, seemed to see the possibility of mechanical reasoning devices using
rules of logic to settle disputes. Both Leibniz and Braise Pascal designed calculating machines
that mechanized arithmetic, which had hitherto been the area, learned men called “calculators,”
but they never made the claim that the devices could think. Etienne Bonnet, Abbey de Cadillac
used the metaphor of a statue in to whose head we poured nuggets of knowledge, asking at what
point it would know enough to appear to be intelligent.
Page 4
Artificial Intelligence
Robots and artificially shaped beings such as the Golem in Jewish tradition and Mary
Shelly’s Frankenstein have always captured the public’s imagination, in part by playing on our
fears. Mechanical animals and dolls-including a mechanical trumpeter for which Ludwig van
Beethoven wrote a trumpet blast- were actually built from clockwork mechanisms in the
seventeenth century. Although they were apparently limited in their performance and were
intended more as curiosities than as demonstrations of thinking, they provided some initial
credibility to mechanistic views of actions and to the idea that such actions need not be feared.
As the industrial world became more mechanized, machinery became classier and more
humdrum. But it was still essentially clockwork. Chess is quite obviously an enterprise that
requires thought. It is not too surprising, then, that chess-playing machines of the eighteenth and
nineteenth centuries, most notably “the Turk,” were exhibited as intelligent machines and even
fooled some people into believing the machines were playing by you. Chess was widely used as
a vehicle for studying supposition and depiction mechanisms in the early decades of AI work.
3.0 Problems:
The general problem has been broken down in to a number of sub-problems. The
following are the different problems
3.1 Learning:
Natural language processing gives machines the ability to read and understand the
language that humans speak. Unsupervised learning is the ability to find patterns in a stream of
input. Supervised learning includes both classification and numerical regression. Classification is
used to determine what category something belongs in, after seeing a number of examples of
things several categories. Regression takes a set of numerical output/input.
Page 5
Artificial Intelligence
3.2 Motion and manipulation:
Robotics field is related to AI. Intelligence is required for robots to be able to grip such
errands as object management and steering.
3.3 Deduction:
Human beings determine most of their problems using fast, impulsive judgments rather
than the conscious, step-by-step assumption that early was able to model. For difficult problems,
most of these algorithms can need huge computational funds the amount of memory or computer
time requires becomes sky-high when the problem goes beyond a positive size.
3.4 Reasoning:
AI has made some progress at imitating this kind of “sub-symbolic” problem solving.
alive agent approaches lay emphasis on the importance sensor motor skills to higher reasoning.
3.5 General intelligence:
Many of the problems above are considered complete to solve one problem, you must
solve them all. Combining all the skills above and exceeding human abilities at most or all of
them, a few believe that anthropomorphic features like consciousness are an artificial brain may
be required for such a project. Even a basic, specific task like machine translation requires that
the machine follow the author’s argument, know what is being talked about and faithfully
reproduce the attention.
Page 6
Artificial Intelligence
3.6 Planning:
They need away to dream of the future and be able to make choices that maximize the
effectiveness of the available choices. Planning must be capable to set goals and achieve them.
In classical problem solving, the agent can assume that it is the only thing acting on the
world and it can be certain what the penalty of it action may be.
3.7 Social intelligence:
Two roles of an intelligent agent are emotion and social skills play.
a) It must be able to predict the actions of others, by understanding their motives and emotional
stats.
b) For good human-computer communication, an intelligent machine also needs to display
emotions. At very least it must appear polite and sensitive to the human it interacts with.
4.0 Approaches;
A few of the most long standing questions that have remained unanswered should AI
simulate neutral intelligence, by studying psychology biology as irrelevant to AI research as bird
biology . Can intelligent behavior to described using simple, elegant principles.
4.1 Integrating the approaches
An agent that solves a specific problem can use any approach that works – some agents
are symbolic and logical, some are sub-symbolic neutral networks and other may use new
approaches. Intelligent agent paradigm is a system that perceives its environment and takes
actions which maximizes its chances of success.
Page 7
Artificial Intelligence
4.2 Brain simulation:
A Manu of the researchers gathered for meetings to the teleological society approach was
largely abandoned, although elements of it would be revived. A system with both symbolic and
sub symbolic components is a hybrid intelligent system, and the study of such system is artificial
intelligence systems integration.
4.3 cognitive architectures:
The connection between the neurology, information theory, and cybernetics,
some of them built machines that used electronic networks to exhibit simple intelligence.
A hierarchical control system provides a bridge between sub symbolic AI at its lowest, reactive
levels and traditional symbolic AI at its highest levels, where relaxed time constraint permit and
world modeling.
4.4 Knowledge based:
The knowledge evolution led to the development and consumption of expert systems the
first truly successful form of Ai software. When computers with large memories became
available around 1970, researchers from all three traditions began to build knowledge AI
applications. The knowledge revolution was also driven by the consciousness that enormous
amounts of knowledge would be required by many simple AI applications
Page 8
Artificial Intelligence
4.5 Symbolic:
Logic based, machines did not need to simulate human thought, but should instead try to find
essence of abstract reasoning and problem solving, regardless of whether people used the same
algorithms. Cosngnitive simulation studied human problem solving skills and attempted to
finalize them, and their work laid the foundations of the field of artificial intelligence, as well as
cognitive science, operations research and management science.
5.0 Tools:
Tools are most important once to solve the most difficult problems AI has developed
large number of tools. The following are the tools mostly used in
5.1 Probabilistic method for uncertain reasoning
Bayesian networks are a very general tool that can be used for large number of problems
reasoning, learning, planning and perception. Many problems in AI in reasoning, planning,
learning, perception and robotics require the agent to operate with incomplete or uncertain
information. Starting in the late 80s and early 90s, Judea pearl and others championed the use of
methods drawn from probability theory and economics to devise a number of powerful tools to
solve these problems.
Probabilistic algorithms can also be used for filtering, prediction, smoothing and finding
explanations for streams of data, helping perception systems to analyze process that occur over
time.
Page 9
Artificial Intelligence
A key concept from the science of economics is utility a measure how valuable
something to an intelligent agent. Precise mathematical tools have been developed that analyze
how dynamic decision networks, game theory and mechanism design.
5.2 Search and optimization
They may begin with a population of organisms and then allow them to mutate and
recombine, selecting only the fittest to survive each generation. Many problems in AI can be
solved in theory by intelligently searching through many possible solutions. A very different
kind of search came to prominence in the 1990s, based on the mathematical theory of
optimization. Planning algorithms search through trees of goals and sub goals, attempting to find
a path to a target goal, a process called means ends analysis. Reasoning can be reduced to
performing a search.Logical proof can be viewed as searching for a path that leads from premises
to conclusions, where each step is the application of an inference rule. Robotics algorithms for
moving limbs and grasping object use local searches in configuration space. Many learning
algorithms use algorithms based on optimization.
Simple exhaustive searches are rarely sufficient for most real world problems. The
searches grow quickly to astronomical numbers. These algorithms can be visualized as blind hill
climbing. We begin the search t a random point on the landscape, and then, by jumps or steps,
we moving our guess uphill, until we reach the top.
The result is a search that is too slow or never completes. The solution, for many
problems, is to use rules of thumb that eliminate choices that are unlikely to lead to the goal.
Heuristics supply the program with a “best guess” for what path the solution lies on.
Page 10
Artificial Intelligence
For many problems, it is possible to begin the search with some form of a guess and then
refine the guess incrementally until no more refinements can be made. Other optimization
algorithms are simulated annealing, beam search and random optimization in the field of study of
the artificial intelligence.
Evolutionary computation uses a form of optimization search. Form of evolutionary
computation includes swarm intelligence algorithms and evolutionary algorithms.
5.3 Logic
Logic was introduced into AI research by john McCarthy in his 1958 advice taker
proposal. Logic is used for information representation and problem solving, but it can be applied
to other problems as well in programming.
Several different forms of logic are used in AI research. Propositional or sentential logic is the
logic of statements which can be true or false. First order logic also allows the use of quantifiers
and predicates, and can express facts about objects, their properties, and their relations with each
other. Fuzzy logic is a version of first order logic which allows the truth of a statement to be
presented as a value between 0 and 1, rather than simply true or false. Fuzzy systems can be used
for uncertain reasoning and have been widely used in modern industrial and consumer product
control systems. Subjective logic models uncertainty in a different and more explicit manner than
fuzzy-logic. A given binomial opinion satisfies within a beta distribution. By this method,
ignorance can be distinguished from probabilistic statements that an agent makes with high
confidence.
evasion logics, non monotonic logics and circumscription are forms of logic designed to
help with evasion reasoning and the qualification problem. Several extensions of logic have been
Page 11
Artificial Intelligence
design to handle specific domains of knowledge, such as explanation logics, situation calculus
and fluent calculus, casual calculus, and modal logic.
5.4 Classifiers and statistical learning methods
A classifier can be trained in various ways there are many statistical and machine
learning approaches. The simplest AI applications can be divided into two types’ classifiers and
controllers. Controllers do however also classify conditions before inferring actions, and
therefore classification forms a central part of many AI systems. Classifiers are functions that use
pattern matching to determine a closest match. They can be tuned according to examples, making
them very attractive for use in AI. These examples are known as observations or patterns. In
supervised learning, each pattern belongs to a certain received, that observation is classified.
The most widely used clarifies are the neural network, kernel methods such as the
support vector machine, k-nearest neighbor algorithm, Gaussian mixture model, immature Bayes
classifier, and decision tree. Classifier performance depends greatly on the characteristics of the
data to be classified. There is no single classifier that works best on all given problems this is
also referred to as the no free lunch theorem. Determining a suitable classifier for problem is still
more an art than science. The performance of these classifiers has been compared over a wide
range of tasks.
5.5 Evaluating progress
A quite different approach measures machine intelligence through tests which are
developed from mathematical definitions of intelligence. The main categories of networks are
acyclic or feed forward neural networks and recurrent neural networks. Among the most popular
feed forward networks are perceptions, multi-layer perceptions and radial basis networks.
Page 12
Artificial Intelligence
Neutral networks can be applied to the problem of intelligent control or learning, using such
techniques as Hebbian learning and competitive learning.
The broad classes of outcome for an AI test are:

Sub-human: Perform worse than most humans

Strong super-human: Performs better than all humans

Super-human: performs better than most humans

Optimal: it is not possible to perform better
This procedure allows almost all the major problems of artificial intelligence to be tested.
However, it is a very difficult challenge and at presents all agents fail.
6.0 Applications
Artificial intelligence has successfully been used in a wide range of fields including medical
diagnosis, stock trading, robot control, law, scientific discovery, video games, toys, and web
search engines. Frequently when a technique reaches mainstream use, it is no longer considered
artificial intelligence, sometimes described as the AI effect. It may also become integrated into
artificial life.
6.1 Game playing
You can buy machines that can play master level chess for a few hundred dollars. There
is some AI in them, but they play well against people mainly through monster force computation
looking at hundreds of thousands of positions. To beat a world champion by monster force and
known unswerving heuristics requires being able to look at 200 million positions per second.
Page 13
Artificial Intelligence
6.2 Speech recognition
In the 1990s, computer speech recognition reached a practical level for limited purposes.
Thus a united airline has replaced its keyboard tree for flight numbers and city names. It is
convenient. On the other hand, while it is possible to instruct some computers using speech, most
users have gone back to the keyboard and the mouse as still more convenient.
6.3 Understanding natural language
Just getting a sequence of words into a computer is not enough. Parsing sentences is not enough
either. The computer has to be provided with an understanding of the domain the text is about,
and this is presently possible only for very limited domains.
6.4 Computer vision
The world is composed of three-dimensional objects, but the inputs to the main human
eye and computers cameras are two dimensional. Some useful programs can work solely in two
dimensions, but full computer vision requires partial three-dimensional information that is not
just a set of two- dimensional views. At present there are only limited ways of representing
three-dimensional information directly, and they are not as good as what humans evidently use.
6.5 Expert systems
A knowledge engineer interviews experts in a certain domain and tries to embody their
knowledge in a computer program for carrying out some task. How well this works depends on
whether the intellectual mechanisms required for the task are within the present state of AI.
When this turned out not to be so, there were many disappointing results. One of the first expert
systems was MYCIN in 1974, which diagnosed bacterial infections of the blood and suggested
Page 14
Artificial Intelligence
treatments. It did better than medical students or practicing doctors, provided its limitations were
observed. Namely, its ontology included bacteria, symptoms, and treatments and did not include
patients, doctors, death, recovery, etc., it is clear that the knowledge engineers forced what the
experts told them in to a predetermined frame work. In the present state of AI, this has to be true.
The usefulness of current expert systems depends on their users having common sense.
6.6 Heuristic classification
One of the most feasible kinds of expert system given the present knowledge of AI is to
put some information in one of a fixed set of categories using several sources of information. An
example is advising weather to accept a proposed credit card purchase. Information is available
about the owner of the credit card, his record of payment and also about the item he is buying
and about the establishment from which he is buying it.
7.0 Uses:
Some computer programs that are used to perform AI tasks are designed to manipulate
symbolic information at extremely high speeds, in order to compensate for their partial lack of
human knowledge and selectivity. Such programs are usually called “expert systems”. The
software systems of this type that have been produced so far are limited in their vocabulary and
knowledge to specific, narrowly defined subject areas.
Knowledge-based expert systems enable computers to make decisions for solving
complicated nonnumeric problems. These expert systems consist of hundreds or thousands of
“if-then” logic rules formulated with knowledge gleaned from leading authorities in a given
field. Programs have also been developed that enable computers to comprehend commands in a
natural language ordinary English. They contain large amounts of information about the meaning
Page 15
Artificial Intelligence
of words pertaining to that subject, as well as information about grammatical rules and common
violations of those rules.
Other programs are designed to simulate human capabilities for problem solving through
the use of highly selective search and recognition methods, rather than through superhuman
processing speeds.
Both experts systems and programs simulating human methods have attained the
performance levels of human experts and professionals in performing certain specific tasks, but
by the mid 1990s there were still no programs that could match human flexibility over wider
domains or in tasks requiring much everyday knowledge.
The ability to identify graphic patterns or images is associated with artificial
intelligence, since it involves both cognition and abstraction. In a system designed with this
capability, a device linked to a computer scans, senses, and transforms images in to digital
patterns, which in turn are compared with patterns stored in the computer’s memory.
The stored patterns can represent geometric shapes and forms that the computer has been
programmed to identify. The computer processes the incoming patterns in rapid succession,
isolating relevant features, filtering out unwanted signals, and adding to its memory any new
patterns that deviate beyond a specified threshold from the old and are thus perceived as new
entities in the area.
8.0 Future:
Major and continuing advances in computer processing speeds and memory sizes have
facilitated the development of AI programs. Although most AI programs attempting to simulate
Page 16
Artificial Intelligence
higher mental functions incorporate the restricted access of limited short term memory, which
restricts humans to carrying out one or a few mental tasks at a time, many investigators have
begun to explore how the intelligence of computer programs can be enhanced by incorporating
parallel processing, the instantaneous execution of several separate operations by means of
computer memories that allow many process to be carried out at once.
The question of which portions of the human brain operate serially and which operate in
parallel has been a topic of intense debate by researchers in both the cognitive sciences and AI,
but no clear verdict had been reached by the mid 1990s.
AI research has thus focused on understanding the mechanisms involved in human
mental tasks and on deceitful software that performs similarly, starting with relatively simple
ones and continually progressing to levels of greater involvedness.
Even as AI technology becomes integrated into the fabric of everyday life, AI researchers
remain focused on the grand challenges of automating intelligence Work is progressing on
developing systems that converse in natural language, that perceive and respond to their
surroundings, and that encode and provide useful access to all of human knowledge and
expertise.
Understanding of human intelligence, abstract understanding of intelligence continues to
have practical consequences in the form of new industries, enhanced functionality for existing
systems, increased productivity in general, and improvements in the quality of life. But the
ultimate promises of AI are still decades away, and the necessary advances in knowledge and
technology will require a sustained fundamental research effort.
Page 17
Artificial Intelligence
REFERENCES:
Bruce G, Buchanan. (Winter 2005) “ A (Very) Brief History Artificial intelligence”

Shapiro, Stuart C. (1992) “Artificial intelligence”

Serenko, Alexander; Deltor, Brain (2004) “Intelligent agents as innovations” AI and
society.

Bernard, Ettienne (2006) “Data characteristics that determine classifier performance”

Norvig, peter (2003), Artificial Intelligence: A modern approach, Upper saddle River,
New Jersey: Prentice Hall, ISBN 0-13-790395-2
Page 18