A. Krone

Technische Universität Dortmund, Dortmund, North Rhine-Westphalia, Germany

Are you A. Krone?

Claim your profile

Publications (13)1.75 Total impact

  • Data mining and computational intelligence; 03/2001
  • Source
    Angelika Krone, Heike Taeger
    [Show abstract] [Hide abstract]
    ABSTRACT: In the field of fuzzy modelling, the exclusive consideration of the modelling error leads to problems concerning the handling of high-dimensional applications and the interpretability of the resulting rule base. To solve those problems, a statistically motivated fuzzy rule test is proposed. It decides if a fuzzy IF/THEN statement is a relevant rule or not. In this way, the problem of finding a good rule base can be reduced to the problem of finding good, relevant rules.
    Fuzzy Sets and Systems 01/2001; · 1.75 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: In this paper we propose a hybrid fuzzy-evolutionary system for fuzzy modelling in high dimensional spaces. The system architecture is based on a Michigan-style approach (one individual represents one fuzzy rule). The design of the evolutionary algorithm makes use of a distance measure in the search space that in turn reflects some heuristic assumptions about the fitness landscape. Additionally, strategy parameters are dynamically adapted by means of a fuzzy controller. The approach is successfully applied to a complex benchmark problem as well as to several real-world modelling tasks such as the cancellation behaviour of insurance clients and the classification of automatic gearboxes.
    05/2000;
  • Source
    P. Krause, A. Krone, T. Slawinski
    [Show abstract] [Hide abstract]
    ABSTRACT: One approach for system identication among many others is the fuzzy identication approach. The advantage of this approach compared to other analytical approaches is, that it is not necessary to make an assumption for the model to be used for the identication. In addition, the fuzzy approach can handle nonlinearities easier than analytical approaches. The Fuzzy{ROSA method is a method for data{based generation of fuzzy rules. This is the rst step of a two step identication process. The second step is the optimization of the remaining free parameters, i.e. the composition of the rule base and the linguistic terms, to further improve the quality of the model and obtain small interpretable rule bases. In this paper, a new evolutionary strategy for the optimization of the linguistic terms of the output variable is presented. The eectiveness of the two step fuzzy identication is demonstrated on the benchmark problem 'kin dataset' of the Delve dataset repository and the results are compared to analytical and neural network approaches.
    04/2000;
  • A. Krone, P. Krause, T. Slawinski
    [Show abstract] [Hide abstract]
    ABSTRACT: In the field of fuzzy modeling, fuzzy control and fuzzy classification, the transparency and comprehensibility of a rule base depends essentially on two aspects: the local relevance of the individual rules and the compactness of the rule base with respect to the number of rules. Additionally, in most applications a certain quality of the input/output behavior has to be achieved. This leads to a multicriteria optimization problem and the global optimum can only be reached for small problem sizes. A possible solution to this problem is a two-step approach. In the first step, a set of relevant rules is incrementally collected. In the second step, the number of relevant rules is reduced to a rule base as small as possible. We present an optimizing conflict reduction, based on a genetic algorithm with bottom-up initialization. The aim is to reach a suitable compromise between the modeling error and the number of rules. Our two-step approach does not aim at locating the global optimum, but at finding a satisfying solution in an acceptable time, even in high dimensional search spaces. In contrast to known approaches, this approach allows rule bases with several ten thousand rules to be handled successfully. The results are illustrated for a benchmark problem and are compared with results of other rule reduction methods
    Fuzzy Systems, 2000. FUZZ IEEE 2000. The Ninth IEEE International Conference on; 02/2000
  • [Show abstract] [Hide abstract]
    ABSTRACT: We propose a hybrid fuzzy-evolutionary system for fuzzy modelling in high dimensional search spaces. The system architecture is based on a Michigan-style approach (one individual represents one fuzzy rule). The design of the evolutionary algorithm makes use of a distance measure in the search space that in turn reflects some heuristic assumptions about the fitness landscape. Additionally, strategy parameters are dynamically adapted by means of a fuzzy controller. The approach is successfully applied to a complex benchmark problem as well as to several real-world modelling tasks such as the cancellation behaviour of insurance clients and the classification of automatic gearboxes
    Fuzzy Systems Conference Proceedings, 1999. FUZZ-IEEE '99. 1999 IEEE International; 02/1999
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Zusammenfassung Im vorliegenden Beitrag wird ein hybrides evolutionäres Suchkonzept zur da-tenbasierten Fuzzy–Modellierung in hoch dimensionalen Suchräumen vorge-stellt. Ausgangspunkt ist ein evolutionärer Algorithmus mit einer Population aus einzelnen Fuzzy–Regeln, die bezüglich ihrer Relevanz evaluiert werden (Michigan–Ansatz). Ziel ist es, trotz der zum Teil extrem unterschiedlichen Anforderungen der Lerndaten, ein gutes und robustes Suchverhalten des evo-lutionären Algorithmus auf verschiedenen Beispielproblemen zu erreichen und insbesondere eine vorzeitige Stagnation der Suche zu vermeiden. In einem hybriden Ansatz wurde hierzu ein Fuzzy Logic Controller zur dy-namischen Adaption der Strategieparameter des evolutionären Algorithmus entwickelt und eingesetzt. Dafür war es zunächst notwendig, geeignete Indi-katoren einzuführen, anhand derer der bisherige Suchverlauf beurteilt werden kann. Darauf aufsetzend wurden Strategien zur Adaption der einzelnen Pa-rameter der evolutionären Regelsuche entwickelt und erprobt. Abschließend wurde anhand sechs ausgewählter Benchmarkprobleme und Anwendungen das Gesamtverhalten des hybriden Suchkonzepts untersucht und ein Leistungsver-gleich durchgeführt.
    01/1999;
  • Source
    A. Krone, H. Taeger
    [Show abstract] [Hide abstract]
    ABSTRACT: this paper: an algorithmic extension of the crisp relevance test, a Bootstrap fuzzy relevance test and an asymptotic fuzzy relevance test. The results of the Bootstrap fuzzy relevance test are very good, but the high computing time makes its application only practicable for a small number of data samples. The asymptotic fuzzy relevance test supplies good results for a higher number of data samples. The algorithmic extension of the crisp relevance test tends to calculate too large condence intervals, but has the smallest computing time. The employment of the three relevance tests will depend on the respective application. For high dimensional search spaces with a multitude of relevant rules, the algorithmic extension is acceptable, especially, if for each input and output variable several trapezium fuzzy sets are reasonable. In the other cases, the higher eoeort of the fuzzy relevance tests can remunerate. The calculation of estimators and condence intervals on fuzzy data is also meaningful for other test and rating strategies, e.g. the results can be directly used for the method 'Condent Normalized Hit Rate' of Jessen and Slawinski [9]. Acknowledgement
    11/1998;
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: This paper presents two applications of the WINROSA software tool. In the first application a data--based generated fuzzy modul is used to adapt the parameters of the position controller of an industrial robot to optimise the continuous path accuracy. It is shown how to learn from good and poor control strategies. The second application is the classification of automatic gear boxes by 149 characteristics. It is demonstrated that a data--based generated fuzzy modul is a promising approach for handling this very complex problem. A new method for complexity reduction is used to reduce the number of necessary process characteristics by analysing their relevance for the classification. 1 Introduction to WINROSA The application of fuzzy models and fuzzy controllers depends on efficient data based methods as direct design by experts is often not possible. The Fuzzy--ROSA 2 method [Krone94] is a concept for generating a fuzzy rule--based model from the input--output behaviour of a d...
    10/1998;
  • A. Krone, T. Slawinski
    [Show abstract] [Hide abstract]
    ABSTRACT: For complex problems of data-based fuzzy modelling the computing time plays an important role. Thus, reduction of the problem size by restricting the search to promising possibilities is justified. This paper presents a new method for extracting unidimensional fuzzy sets from measured data for a subsequent rule generation process. This method is motivated by four main points: 1) the projection of multidimensional data to unidimensional fuzzy sets considers the dependence between the input variables and the output variable without anticipating the rule generation process; 2) the user is not required to predefine the number of fuzzy sets and the number is changeable in a flexible manner for each variable without new computations; 3) the sum of membership values of one variable is one; and 4) the computing time does not increase more than linearly with the number of input variables
    Fuzzy Systems Proceedings, 1998. IEEE World Congress on Computational Intelligence., The 1998 IEEE International Conference on; 06/1998
  • A. Krone, U. Schwane
    [Show abstract] [Hide abstract]
    ABSTRACT: The design of fuzzy logic controllers for complex processes can be supported by the analysis of available observation data. Two main problems, however, arise in this context. First, in real world domains the data are contradictory. Second, the observation data can differ widely in quality, so that a uniform treatment or a simple classification into good and bad observation data leaves too much valuable information not being considered. In this paper, the fuzzy-ROSA method (Rule Orientated Statistic Analysis) is presented with a new concept for generating significant and quality-orientated fuzzy rules from observation data of different control strategies and control performances of the process under consideration. The method is illustrated by an application to an industrial six-axis robot arm
    Fuzzy Systems, 1996., Proceedings of the Fifth IEEE International Conference on; 10/1996
  • A. Krone, H. Kiendl
    01/1994;
  • Source