See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/220204826

Engineering Intelligent Systems, Introduction

Article in Applied Intelligence · September 2004

DOI: 10.1023/B:APIN.0000033656.71725.23 · Source: DBLP

CITATION READS

1 236

1 author:

Laszlo Monostori
Hungarian Academy of Sciences
92 PUBLICATIONS 4,541 CITATIONS

SEE PROFILE

All content following this page was uploaded by Laszlo Monostori on 10 August 2017.

The user has requested enhancement of the downloaded file.

Monostori, L.: Engineering intelligent systems, Introduction, Applied Intelligence: The
International Journal of Artificial Intelligence, Neural Networks, and Complex Problem-Solving
Technologies, Kluwer Academic Publishers, Boston, Dordrecht, London, Vol. 21, No. 2,
September-October 2004, pp. 115-116.
Special Issue on Engineering Intelligent Systems

Introduction

The success of engineering work depends on the recognition and use of several synergic
factors. While the analytic approach is the key in solving some well-defined subproblems,
creativity helps us to get free of prejudices and in finding useful analogies. Moreover, critical
thinking opens up the way of argumentation with inexact premises and rules. While
engineering is based on complex mental features like this, it is perhaps more transparent than
some other human activities: its world is that of man-made artifacts that should be useful in a
broad sense of the word.

That is why we think that by considering the advance in Engineering of Intelligent Systems,
one may have a proper view of the state of research in the whole of Artificial Intelligence
(AI). Lessons learnt here are valid not only in this particular application area but give useful
hints to the perspectives in the theory of AI and its potential for solving other problems, as
well.

Having recognized the above features of engineering problem solving, the International
Society of Applied Intelligence (ISAI), more than a decade ago, initiated a series of
conferences named Industrial & Engineering Applications of Artificial Intelligence & Expert
Systems (IEA/AIE). The Hungarian AI community was honored by the invitation of the ISAI
to organize and host the IEA/AIE-2001 conference (June 4-7, 2001, Budapest, Hungary),
which was the fourteenth one in the series.

The Editor faced a difficult problem: how to select the best papers from the versatile – both
from the views of the AI techniques and application fields – a palette of 104 presented papers
arranged in conference sections are as follows: Search, Knowledge Representation, Model-
Based Reasoning, Machine Learning, Data Mining, Soft Computing, Evolutionary
Algorithms, Distributed Problem Solving, Expert Systems, Pattern and Speech Recognition,
Vision, Language Processing, Planning and Scheduling, Robotics, Autonomous Agents,
Design, Control, Manufacturing Systems, Finance and Business, Software Engineering, and
Tutoring.

Finally, the authors of 6 papers have been asked to present the in-depth-versions of their
conference papers in this special issue.

The paper of Y.L. Murphey, H. Guo and L.A. Feldkamp deals with the problem of unbalanced
and noisy data in course of neural learning. Capabilities of different neural network
architectures and training algorithms are compared by using data examples downloaded
directly from test sites of an automobile assembly plant. The authors underline that by
properly generated extra training data examples around the noise densities, neural networks
with increased generalization capabilities can be generated.

A. Kocsor and L. Tóth describe the main results of their investigations concerning
classification and transformation methods applied to phoneme classification. The achieved
scores are compared with the performance of a traditional hidden Markov phoneme model
Monostori, L.: Engineering intelligent systems, Introduction, Applied Intelligence: The
International Journal of Artificial Intelligence, Neural Networks, and Complex Problem-Solving
Technologies, Kluwer Academic Publishers, Boston, Dordrecht, London, Vol. 21, No. 2,
September-October 2004, pp. 115-116.
(HMM). The paper is expected to be a useful starting point for those working in the described
domain, but some of the results can be transferred to other application fields, as well.

Aiming at real-time applications, the efficient hardware implementation of neural networks is

of fundamental importance. In their paper, T. Szabó and G. Horváth propose a new way of the
digital hardware implementation of nonlinear activation functions in feed-forward neural
networks, by using matrix-vector multiplications. In the described solution the nonlinear
sigmoidal functions are approximated by B-spline basis functions. The results of an extensive
experimental study show that the drawbacks of the approximation can be easily compensated
by a slightly larger number of hidden units in the network.

F. Wotawa addresses the problem of software debugging of large programs, usually written
by different groups of programmers. The problem is strongly manifested in the domain of
hardware design where special hardware description languages, such as VHDL is used.
Model-based diagnosis is proposed as a general diagnosis approach, with models based on the
logical descriptions of the syntax and semantics of a VHDL program and can be automatically
derived from the program at compile time.

The evergreen theme of scheduling is treated in the paper by M.I. Alfonso and F. Barber who
propose a new method that efficiently integrates the CSP (Constraint Satisfaction Problem)
process into a limited closure process: not only interleaving them but rather as a part of the
same process. The described experiments with flow-shop and job-shop instances show that by
using their approach, feasible/optimal solutions can be obtained without having to use
backtracking in most cases.

Information extraction (IE) from the web is a hot topic these days. L. Xiao, D. Wissmann, M.
Brown and S. Jablonski illustrate how real world applications and different domains can be
addressed by autonomous, corporate components (agents). As a further point, the applicability
of machine learning in practical IE systems is shown. Different learning approaches are
proposed at different levels of the systems.

The Guest Editor would like to thank all the authors for their valuable contributions to this
special issue. Our thanks are due to the International Society of Applied Intelligence (ISAI)
that acted as the main sponsor of the 14th IEA/AIE Conference, and also to the co-sponsors:
• the American Association for Artificial Intelligence (AAAI),
• the Association for Computing Machinery (ACM/SIGART),
• the Canadian Society for Computational Studies of Intelligence (CSCSI),
• the European Research Consortium for Informatics and Mathematics (ERCIM),
• the Hungarian Academy of Sciences (HAS),
• the Institution of Electrical Engineers (IEE),
• the International Neural Network Society (INNS),
• the Japanese Society of Artificial Intelligence (JSAI),
• the Ministry of Education, Hungary (OM), and
• the Southwest Texas State University (SWT).
We thank the members of the International Program Committee, especially for those of a
specific role: Dr. József Váncza (Local Chair), Prof. Soundar Kumara (Program Co-Chair),
Prof. Khosrow Kaikhah (Publicity Chair) and Dr. Gusztáv Hencsey (Local Organization
Chair). We would also like to thank all the authors, and referees for their contribution and
efforts which made IEA/AIE-2001 possible.
 Monostori, L.: Engineering intelligent systems, Introduction, Applied Intelligence: The
International Journal of Artificial Intelligence, Neural Networks, and Complex Problem-Solving
Technologies, Kluwer Academic Publishers, Boston, Dordrecht, London, Vol. 21, No. 2,
September-October 2004, pp. 115-116.
I wish to express my personal gratitude to the Editor-in-Chief for his initiative,
encouragement and continuous support.

László Monostori
Computer and Automation Research Institute, Hungarian Academy of Sciences
Budapest University of Technology and Economics
[email protected]

View publication stats