Christina Klüver

Self-organized learning by self-enforcing networks

We describe a new type of self-organized learning neural networks, namely the Self-Enforcing Network SEN. After introducing our theoretical and methodical frame, basically orientated to the theory of Piaget, we show the logical principles of SEN, including a new activation function, a new learning rule, and a specific visualization algorithm. The operations of SEN are demonstrated with the examples of assimilating new perceptions of animals, the application of a SEN to direct marketing, and the usage of a SEN as consulting system for pupils and beginners at the university. Apparently SEN can be used in many different contexts.

The regulatory algorithm (RGA): a two-dimensional extension of evolutionary algorithms

In orientation to new developments in evolutionary biology we propose an extension of evolutionary algorithms in two dimensions, the regulatory algorithm (RGA). It consists of two levels of vectors, the regulatory vector and the structural vector. Each element of the regulatory vector is connected with one or several elements of the structural vector, but not vice versa. The connections can be interpreted as steering connections, the switching on or off of the structural elements and/or as switching orders for the structural elements. An RGA that operates with the usual genetic operators of mutation and crossover can be used for avoiding rules like penalty or default operators, it is in certain problems significantly faster than a standard genetic algorithm, and it is very suited when modeling and optimizing systems that consist themselves of different levels. Examples of RGA usage are shown, namely, the optimal dividing of socially deviant youths in a hostel, the optimal introduction of communication standards in information systems, and the allocation of employees to superiors by taking into regard the different personality types.

Steering clustering of medical data in a Self-Enforcing Network (SEN) with a cue validity factor

A Self-Enforcing Network (SEN), which is a self-organized neural network, is introduced to cluster medical data. In addition, a cue validity factor is defined, which affects the clustering of the data. The results show that a user can influence the clustering of data by SEN, thus allowing the analysis of the data depending on economical, medical or nursing interests. The described prototype includes concrete examples and shows the potential of such a network for the analysis of complex data.

Solving problems of project management with a self enforcing network (SEN)

The usage of AI techniques when dealing with problems of management always implies the task to understand the respective problem in terms of these techniques, in this case in terms of a specific neural network. That is, by the way, at its core a hermeneutical problem, namely of “understanding” the problem situation in the sense of hermeneutics. I shall demonstrate how a specific self-organized learning network that we have newly developed can be applied to problems of project management. This new network SEN will be described and its possibilities are shown by the application to different problems of project management, e.g. the selection of suited collaborators for a specific project, the classification of problematic customers or also the selection of suited procedures for project management in a specific firm. The SEN seems to be a universal instrument for different purposes as many reactions from managers in several large firms have shown. In particular SEN seems to be more suited for practical problems than some standard software we compared with SEN.

A self-enforcing neural network as decision support system for air traffic control based on probabilistic weather forecasts

We demonstrate the application of a new self-organized learning neural network, the Self-Enforcing Network SEN, to the problem of selecting the right direction of operation for runways at the airport of Frankfurt/M. The SEN is given real data on different days for the forecasted wind situations and generates a recommendation for the selection of a suited direction of operation. The results demonstrate that the recommendations of SEN are sound ones insofar that they are very compatible with the factual decisions of air traffic control at Frankfurt Airport. In some cases air traffic control could have planned earlier with the results of SEN. Hence this SEN system demonstrates that it could be a helpful decision support system for this problem.

Optimization of room allocation plans at the university Duisburg-Essen with a regulatory algorithm

We describe a new evolutionary algorithm, namely the regulatory algorithm (RGA), a two dimensional extension of standard evolutionary algorithms. Its possibilities are shown by an application to the problem of optimizing room allocation plans with real data from the University Duisburg-Essen (Germany). The results of the RGA application match in most cases the factual room allocations of the university and are in several cases better, i.e. the RGA results fulfill more necessary conditions.

Word morph and topological structures: A graph generating algorithm

An algorithm of generating graph structures for the ordering of datasets is proposed. It operates with the key concept of ‘‘sphere neighborhood’’ and generates the graph structured ordering of datasets, provided the elements of such sets can be related by some similarity relation. The algorithm always allows determining the shortest path between two nodes in a constructive way. It is demonstrated by an application to the famous Word Morph game and in addition to the ordering of logfiles with respect to the detection of errors. Therefore, the algorithm can be very useful for the dealing with Big Data problems. VC 2016 Wiley Periodicals, Inc. Complexity 000: 00–00, 2016

OSWI: a consulting system for pupils and prospective students on the basis of neural networks

The consulting system OSWI—Orientating System for Economy and Computer Science—has been developed by commission of the Faculty of Economy and Computer Science of the University of Duisburg-Essen. The system is based on the characteristics and expected abilities for the different courses of study of the faculty, which were formulated by the professors as representatives of their respective fields. The algorithmic base of the system is a new self-organized learning neural network that has been developed by us, namely the self-enforcing network. OSWI has been used, at present, by more than 5,000 students and pupils and was always evaluated by the users as a very useful and user-friendly system for selecting among the faculty’s courses. The methodical approach for obtaining the database of the system and the algorithm with which OSWI operates is both highly innovative and very well suited for other areas.

A Self-Enforcing Network as a Tool for Clustering and Analyzing Complex Data

Abstract The Self-Enforcing Network (SEN), which is a self-organized learning neural network, is introduced as a tool for clustering to define reference types in complex data. In order to achieve this, a cue validity factor is defined, which first steers the clustering of the data. Finding reference types allows the analysis and classification of new data. The results show that a user can influence the clustering of data by sEN, thus allowing the analysis of the data depending on specific interests. The described tool includes concrete examples with real clinical data and shows the potential of such a network for the analysis of complex data.

Finding optimized configurations for variability-intensive systems without constraint violations using a Regulatory Algorithm (RGA)

Feature selection is one of the important challenges in variability-intensive systems. The FCORE model is used for the description of the functional and non-functional requirements of a system from a systems engineering point of view. In addition we demonstrate a solution for feature selection using a regulator algorithm (RGA). The RGA is a two dimensional evolutionary algorithm, with regulator genes controlling the structural genes. This allows a direct transfer of the FCORE model into the RGA, which optimizes the feature selection without constraint violations.