Int. J. of Computers, Communications & Control, ISSN 1841-9836, E-ISSN 1841-9844
Vol. III (2008), Suppl. issue: Proceedings of ICCCC 2008, pp. 28-32
World Knowledge for Control Applications by Fuzzy-Interpolative
Systems
Parallel session invited paper
Marius M. B˘ ala¸ s, Valentina E. B˘ ala¸ s
Abstract: The paper is discussing the necessity of providing controllers with incipient el-
ements of world knowledge: general knowledge on system theory, specific knowledge on
the processes, etc. This can be done by means of the fuzzy-interpolative systems, allied
with simulation models and/or planners. A structure of world knowledge embedding planned
controller is illustrating the idea.
Keywords: fuzzy-interpolative controllers fuzzy-interpolative expert systems, knowledge
embedding by computer models, planners.
1 Introduction
In ref. [1] Lotfi A. Zadeh affirmed that the main weakness of the Question-Answering Systems is the absence
of the world knowledge. World knowledge WK is the knowledge acquired through experience, education and
communication.
The components of WK are [1]:
• Propositional: Paris is the capital of France;
• Conceptual: Climate;
• Ontological: Rainfall is related to climate;
• Existential: A person cannot have more than one father;
• Contextual: Tall.
Some of the main characteristics of WK are:
• Much of WK is perception-based;
• Much of WK is negative, i.e., relates to impossibility or nonexistence;
• Much of WK is expressed in a natural language.
Obviously KW is highly necessary to the human emulating AI products. Nevertheless this approach must
overcome lots of difficulties: WK need huge memory capacity, the representation techniques must be in the same
time comprehensive, specific and portable, the selection of the knowledge, the learning and the forgetting processes
need further fundamental conceptual investigations, etc.
The aim of this paper is to answer the following question: "Can low level computing devices: μ P, μ C, DSP,
etc. benefit of WK, when even the sophisticated modern AI software, running on powerful workstations, is en-
countering difficulties?"
2 The Fuzzy-Interpolative Systems
A fuzzy-interpolative controller FIC is a fuzzy controller that can be equaled with a corresponding look-up table
with linear interpolations. The FIC concept must not be confounded with the fuzzy rule interpolation, originally
introduced by L.T. Kóczy and K. Hirota [2], [3].
A typical FIC is a Sugeno controller with triangular or trapezoidal fuzzy partitions, prod-sum inference and
COG defuzzyfication [4], [5], [6], [7], etc. The interpolative counterpart of this controller is the look-up table with
linear interpolations (as the corresponding Simulink-Matlab block). FICs started from the practical observation that
the Matlab FIS (Fuzzy Inference System) toolkit is demanding notable resources and occasionally it encounters
Copyright © 2006-2008 by CCC Publications - Agora University Ed. House. All rights reserved.