 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
|
|
|
Next issue (December 2002) Security in e-Commerce |
 |
October 2002 Artificial Intelligence:
|
Guest
Editors:
Federico Barber, Vicente
J. Botti, and Jana Koehler
|
|
Acrobat Reader is required to display PDF files
 |
|
 |
|
 |
|
 |
|
 |
|
 |
|
|
|
Guest
Editors:
Acrobat Reader is required to display PDF files |
|
Presentation. AI: Past,
Present and Future [HTML] [PDF:
4 pages, 648 KB]
(Includes a list of Useful
References for those interested in knowing more about AI)
Federico Barber, Vicente
J. Botti, and Jana Koehler
Abstract: The guest
editors outline the history of Artificial Intelligence, present the issue
and include a list of useful references for those interested in knowing
more about Artificial Intelligence. They explain that AI represents a serious
effort to understand the complexity of human experience in information
processing terms, and deals not only with how to represent and use complex
and incomplete information logically, but also with questions of how to
see (vision), move (robotics), communicate (natural language, speech),
learn, etc.
Spoken Communication with
Computers [PDF: 4 pages, 634 KB]
Francisco Casacuberta-Nolla
Abstract: Present
day technology allows us to build commercial IT systems able to transcribe
speech into written text, interpret a spoken instruction in order to manage
devices, or access information systems by means of a dialogue between human
being and machine. The same technology also allows us to build systems
which translate speech from one language into another for limited tasks.
The great success of these systems is due, among other reasons, to the
use of pattern recognition techniques and, in particular, to the fact that
models can be built automatically from examples of the problem to be tackled.
However, we are still a long way off achieving real spoken communication
between human beings and computers. We may need a new framework in which
to develop new models and techniques.
Progress in AI Planning
Research and Applications [PDF: 15 pages, 748
KB]
Derek Long and Maria
Fox
Abstract: Planning
has made significant progress since its inception in the 1970s, in terms
both of the efficiency and sophistication of its algorithms and representations
and its potential for application to real problems. In this paper we sketch
the foundations of planning as a sub-field of Artificial Intelligence and
the history of its development over the past three decades. Then some of
the recent achievements within the field are discussed and provided some
experimental data demonstrating the progress that has been made in the
application of general planners to realistic and complex problems. The
paper concludes by identifying some of the open issues that remain as important
challenges for future research in planning.
Trends in Automatic Learning
[PDF:
7 pages, 653 KB]
Ramon López de
Mántaras
Abstract: The capacity
for learning is one of the fundamental characteristics of intelligence.
Learning forms part of any activity requiring intelligence, such as diagnosis,
planning, language, motor-sensory activities, etc. For this reason automatic
learning plays a vital role in applications which have proven to be to
difficult to programme manually. To describe each and every method of automatic
learning would require a book of its own and besides, the more classic
methods are described in most introductory texts to Artificial Intelligence.
For this reason, this article concentrates on the more modern techniques
of automatic learning..
Knowledge-Based Systems
[PDF:
7 pages, 666 KB]
José Mira-Mira
and Ana E. Delgado-García
Abstract: In this
article the authors introduce the concept of a Knowledge-Based System (KBS)
as a computer program, which codifies the formal model underlying the knowledge
which a human expert uses to solve a task within a limited field. We also
examine some of its basic characteristics, such as the separation between
domain knowledge, and inference knowledge, which specifies the reasoning
steps followed to solve the task. Then, some methodological issues concerning
the taxonomy of levels and domains of description of our models of human
knowledge are introduced. They also present the different phases in the
development of a KBS (modelling at the knowledge level, “operationalization”
of the inferences, implementation, validation, and evaluation) and end
drawing links with the tendencies of component reuse and integration of
the symbolic and connectionist perspectives of KBS. An essential characteristic
in the perspective of this article is the emphasis on the methodological
aspects, whose aim is to reduce the gap between knowledge engineering and
electronic engineering.
Cooperating Physical Robots
and Robotic Football [PDF: 7 pages, 681 KB]
Bernhard Nebel and Markus
Jäger
Abstract: Having
a robot that carries out a task for you is unquestionably of some help.
Having a group of robots seems to be even better because in this case the
task may be finished that much quicker and more reliably. In spite of this,
dealing with a group of robots can make some problems more difficult. In
this paper the authors sketch some of the advantages and some of the disadvantages
that come up when dealing with groups of robots.
Multi-Agent Systems [PDF:
7 pages, 652 KB]
Carles Sierra
Abstract: Throughout
the last decade there has been a marked increase in the interest shown
by the academic and industrial world in the agent-based programming paradigm,
though many IT engineers are still either unaware of it or confuse it with
the object oriented paradigm. In this article the authorintroduces the
agent-based programming paradigm in some of its many facets and give an
overview of the future lines that research may follow, its main aim is
to encourage readers to go on to read more papers on the subject.
Artificial Intelligence
and Education: an Overview [PDF: 8
pages, 657 KB]
Maite Urretavizcaya-Loinaz
and Isabel Fernández de Castro
Abstract: This paper
presents an overview of various contributions which Artificial Intelligence
has made to the world of computer-aided learning. After a short introduction
to the field, including a brief history of intelligent educational systems,
the authors present some of the pedagogical trends which have influenced
the development of these systems, a non-exhaustive review of intelligent
educational systems, the Artificial Intelligence techniques used in them
and some current lines of research.
Federico Barber, Telecommunications Engineer and Doctor of Computer Science, is currently a Full Professor at the Universidad Politécnica de Valencia (Spain), where has been the Dean of the Faculty of Computer Science. He has been the editor of "Inteligencia Artificial, Revista Iberoamericana de IA"’ (an Ibero-American AI journal), and he is currently President of the Spanish Association for Artificial Intelligence (AEPIA). His areas of study are centred mainly on the problems of constraint satisfaction (scheduling, optimisation, temporal planning with resources, temporal reasoning, etc.) in which he has developed his own models and applications, in addition to the field of knowledge engineering. He is joint leader of an extensive research group and has published a great many scientific articles. He has also participated in or led national and international research projects (CICYT, MC&T, ESPRIT, etc.), and technology transfer agreements, as well as sitting on various scientific committees in his field. He is a senior member of ATI and co-editor of Novática's AI section. <fbarber@dsic.upv.es>
Vicente J. Botti, Electrical Engineer and Doctor in Computer Science, is currently a Full Professor at the Universidad Politécnica de Valencia (Spain), where he has also been the Head of the Dept. of Informatics Systems and Computation. His fields of study are focused mainly on multi-agent systems, and more specifically, real time multi-agent systems, real time systems, mobile robotics (in which he has developed his own models, architectures and applications) in addition to the field of knowledge engineering. He is joint leader of an extensive research group whose general line of research is Artificial Intelligence and has published about 100 scientific articles. He has been and is a principal researcher on nationally and internationally funded projects (CICYT, MC&T, ESPRIT, etc.), and on technology transfer agreements, as well as sitting on various scientific committees in his areas of interest. He is a senior member of ATI and co-editor of Novática's AI section. <vbotti@dsic.upv.es>
Jana Koehler
is a research staff member and project leader at the IBM Research Lab in
Zurich that she joined in Spring 2001. She got her Phd in 1994 from the
University of Saarbruecken, where she had worked at the German Research
Centre for AI from 1990 to 1995 in an AI planning project. From 1996 to
1999 she was an assistant professor at the University of Freiburg where
she started working as a consultant for the technology management of Schindler
Elevators in 1998. From 1999 to 2001 she worked as a project leader for
Schindler Elevators. At IBM, she works on new middleware technology for
the integration and automation of business processes based on webservices.
<koe@zurich.ibm.com>
Artificial Intelligence (AI), defined as “the science of making machines do things that would require intelligence if done by men” (Minsky), took on a viable scientific meaning as a modern Computer Science (CS) discipline during the second half of the 20th century. It was the direct result of the convergence of various intellectual currents (Theory of Computation, Cybernetics, Information Theory, Symbolic Processing, Psychology, …) which had developed from the formal bases of Logic and Discrete Mathematics, and had been given impetus by the development of digital computers. AI represents a serious effort to understand the complexity of human experience in information processing terms. It deals not only with how to represent and use complex and incomplete information logically, but also with questions of how to see (vision), move (robotics), communicate (natural language, speech), learn, etc.
Human intelligent behaviour of the sort that AI tries to emulate comprises several different aspects. One deals mainly with cognitive reasoning processes and is clearly related to logic. Another is more that of a perceptive nature (vision, speech, etc.) and, although it shares some problems and methods with the previous aspect, it tends to be more rigorous in terms of formal expression and its specific problems, techniques and methods constitute the discipline known as Pattern Recognition. Finally we can talk about symbolic AI, concerned with the processing of symbols of knowledge, and connectionist AI, which the process of intelligence is simulated by means of basic, usually quantitive, elements of processing.
If we look at just the common
core of AI, there is a wide range of trends which consider aspects of both
human thought and human behaviour. Each of these trends can in turn receive
empirical approaches which use hypothesis and subsequent confirmation by
experiment, or rationalist approaches which require a combination of logic-mathematical
and engineering processes (see Table 1).
Table 1. Various approaches
to AI from different perspectives
Source: Stuart Russell
and Peter Norvig, Artificial Intelligence: A Modern Approach, Prentice
Hall, 1995
The approaches contained in Table 1 define AI according to each of these different aspects. The definitions in the top part are focused on processes connected with reasoning or thought and the ones in the lower part are focused on processes related to behaviour. The definitions in the left column measure the success of AI from the human perspective (which requires an empirical approach) and those in the right column do the same from a rational perspective, a concept of intelligence which could be called rationality.
In recent years, research into AI has undergone a marked change with regard to both content and the methodology being used. It is becoming more common to build AI systems based on existing theories rather than putting forward newtheories; taking as a starting point rigorous theorems or solid experimental evidence rather than intuition, and demonstrating the use of AI applications in the real world rather than creating ‘toy’ examples. In areas such as games, logical inference and theorem proving, and medical diagnosis, systems based on rigorous theoretical principles are emerging which can perform as well or better than human experts. In other areas – such as learning – visual perception, robotics and natural language understanding are making rapid steps forward thanks to the application of better analytical methods and a better understanding of the problems involved.
A good example of the above is the field of natural language understanding. In the 70s a great many architectures and ad hoc approaches were tested on some specially chosen examples. More recently these have given way to approaches based on hidden Markov models, founded on rigorous mathematical theory in which models are generated by means of a learning process based on a large body of real language data. The use of these models enables us to obtain ever better classifications, and language technology, together with its associated field of handwriting recognition, is currently moving towards industrial and consumer applications.
During the 90s, Fuzzy Logic was consolidated in several AI contexts, and the connectionist paradigm continued to gain favour, as did genetic algorithms, leading to the development of hybrid systems in a quest for adaptability. New knowledge acquisition methodologies were developed, such as KADS (by Stewart Tansley). In learning significant advances were made and new methods were put forward. With regard to cognitive architectures we saw the revolution that the introduction of reactivity sparked in the development of autonomous agents. Finally we witnessed a change of paradigm in artificial vision, from the classic passive approach to the active approach (Alan Yuille) whereby the perceptual task is connected with the performance of actions (perception-action). This had important implications for the development of robotic systems with enhanced performance.
The work done by Tate and Chapman has given rise to an elegant synthesis of planning programmes brought together in a unified framework. Planning systems are currently used for programming the work in factories and for space shots. Meanwhile, intelligent scheduling systems based either on the Constraint Satisfaction Problems (CSP) paradigm extended by the inclusion of temporal reasoning techniques, or knowledge based systems, provide an alternative answer to classic unsolvable problems.
Pearl’s Probabilistic Reasoning in Intelligent Systems (1988) marked the arrival of the use of probability and decision theory in AI. The development of the belief network formalism responded to the need to be able to reason efficiently when faced with a combination of uncertain knowledge. This approach far outperforms the probabilistic reasoning systems of the 60s and 70s, and is currently at the heart of research into AI which is being currently being carried out on uncertain reasoning and expert systems (ES). The work of Pearl, Horvitz and Heckerman served to promote the idea of ES rules, that is, that they should act rationally in accordance with decision theory, without trying to imitate human experts. Following this line of thought, fuzzy logic, based on possibility theory, emerged in response to the difficulty of providing problems with precise data input. Possibility theory was introduced by Zadeh in 1965 to handle uncertainty in fuzzy systems, and has much in common with probability. Although mathematicians at first considered it to be a flawed theory, possibility theory actually tackles a different problem. Fuzzy logic has been widely used by the Japanese in the design and construction of household appliances.
Similar trends have been seen in robotics, computerized vision, automatic learning (including neural networks) and knowledge representation. A better understanding of the problems and their complexity, together with greater computing capacity, have enabled sound reasoning methods to be created. Possibly encouraged by the progress made in the solution of subproblems in AI, researchers have gone back to work on the problem of the “complete agent”, adopting this new, more formalist, trend. Research by Newell, Laird and Rosenbloom (SOAR) is the best known example of a general architecture for an AI system. One of the fundamental aspects of a general architecture is its capacity to incorporate many different kinds of decision making, from knowledge based deliberation to reflex action responses. The new agent architectures aim to strike a balance between these two factors, reflex responses, for situations in the which speed is of the essence, and knowledge based deliberations, where the agent has time to take more information into consideration, for forward planning, for handling situations in which there is no immediate response available and to propose better responses tailored specifically to the situation in hand. Architectures such as SOAR have precisely this structure. By means of compilation processes like explanation based learning, they convert declarative information at a deliberative decision making level into more efficient representations until the decision eventually becomes a reflex action.
Research into real time AI looks into all the above mentioned aspects. Agents in real environments need to have the means of controlling their own deliberations and also be capable of using the time allowed for reasoning to perform the calculations which will provide the best results. As AI systems are applied to ever more complex domains, so all the problems will become real time problems, since the agent will never have enough time to find an exact solution to a problem.
There is obviously a great need for methods which work well in more general decision taking situations. In recent years two promising techniques have appeared, anytime algorithms and decision theory techniques. The last element of an agent’s architecture is its learning mechanism. Inductive learning, reinforcement learning and compilation learning mechanisms can be used for all agent’s learning activities. These mechanisms will doubtless depend on the type of representation chosen. Representations based on logic, and neural and probabilistic networks, are well known and much studied formalisms for which there are a great variety of learning methods. As new representations are created, such as first order probabilistic logics, it will be necessary to create new learning algorithms for them.
Agent/multi-agent system (MAS) technology is making important contributions to problem solving in various domains (e-commerce, e-auctions, medicine, stock market, manufacturing systems, telephony systems, etc.), where traditional approaches do not provide satisfactory solutions. The study of Multi-Agent Systems began nearly 20 years ago, within the area of Distributed Artificial Intelligence (DAI) which is a subfield of artificial intelligence research. DAI is the study of intelligent group behaviour stemming from the cooperation of what are known as agents. It studies how a group of modules cooperate to divide up and share the knowledge of a problem, and how it reaches a solution. DAI focuses on global behaviour, with a predetermined agent behaviour. It studies the techniques and knowledge required for the coordination and distribution of knowledge and actions in a multi-agent environment.
When we look at how AI has evolved in the last fifty years we can see a transition from the initial embryonic theories and systems to the adaptable, robust and user-friendly environments of today; environments based on a wide range of logical theories, cognitive models and engineering based approaches. Technological development and progress in related fields (Neurophysiology, Psychology, Biology) will have a great deal to say in the future. An analysis of current AI systems and the way they can be extended will enable us to pose a great many questions, the answers to which will lead us towards general purpose intelligent systems.
In this monograph by Upgrade we have introduced a few of the areas and techniques involved in AI which, by their very scope, are impossible to deal with comprehensively. We shall, pay special attention to the discipline’s applicability and use as an alternative solution where other techniques or methodologies have failed or do not provide satisfactory solutions, or where these alternative techniques may provide better solutions.
The articles included are the work of very important researchers/developers and cover each of the areas dealt from a multiple viewpoint – generalist, scientific and applied – with a special emphasis on future development. These contributions should give the reader an idea of the historical perspective, the current state and the future possibilities of AI. We hope that they will enable the reader to have a clearer understanding of these areas and a greater awareness of the current realities and the challenges they pose. The articles included are:
“Spoken Communication with Computers”, by Francisco Casacuberta-Nolla, dealing with the development of systems which enable spoken interaction with computers, of widespread use in speech recognition systems, translation systems, etc.
“Intelligent Planning: Progress in Planning Research and Applications”, by Derek Long and Maria Fox, in which they take a look at the applications and current challenges posed by intelligent planning techniques, used in task planning, robots, resource scheduling, etc.
“Trends in Automatic Learning”, by Ramon López de Mántaras, in which he looks at intelligent IT systems’ learning capacity, one of the fundamental characteristics of intelligence, and the techniques they employ to develop it.
“Knowledge-Based Systems”, by José Mira-Mira and Ana E. Delgado-García. In this article Knowledge Engineering is presented with special emphasis on methodological aspects Knowledge Based Systems, Expert Systems), with the aim of approaching the rigour of other engineering disciplines.
“Cooperating Physical Robots and Robotic Football”, by Bernhard Nebel and Markus Jäger. In this article an analysis is made of the techniques and applications related to physical robots in tasks carried out in real environments, where the ability of the robots to cooperate correctly is especially important.
“Autonomous Agents and Multi-Agent Systems”, by Carles Sierra. This article presents the current state of multi-agent systems and their main applications.
“Artificial Intelligence and Education: an Overview”, by Maite Urretavizcaya-Loinaz and Isabel Fernández de Castro. An overview of the different contributions AI is making to the world of educational IT, and a review of intelligent educational systems.
To close, we would like to thank all the participants in this monograph for their interest and efforts, and to thank the editors of Upgrade too for their support, suggestions and infinite patience in bringing this work to fruition.
Note: This monograph
will be also published in Spanish (full issue printed, some articles online)
by Novática, journal of the Spanish CEPIS
society ATI (Asociación
de Técnicos de Informática) at <http://www.ati.es/novatica/>,
and in Italian (online edition only, containing abstracts and some articles)
by the Italian CEPIS society ALSI and the Italian IT portal Tecnoteca at
<http://www.tecnoteca.it>.
| Last updated on October 29th, 2002 | by Rafael Fernández Calvo and François Louis Nicolet |
| <rfcalvo@ati.es> |