Studies in Computational Intelligence

Transcription

1 Studies in Computational Intelligence Volume 758 Series editor Janusz Kacprzyk, Polish Academy of Sciences, Warsaw, Poland

2 The series Studies in Computational Intelligence (SCI) publishes new developments and advances in the various areas of computational intelligence quickly and with a high quality. The intent is to cover the theory, applications, and design methods of computational intelligence, as embedded in the fields of engineering, computer science, physics and life sciences, as well as the methodologies behind them. The series contains monographs, lecture notes and edited volumes in computational intelligence spanning the areas of neural networks, connectionist systems, genetic algorithms, evolutionary computation, artificial intelligence, cellular automata, self-organizing systems, soft computing, fuzzy systems, and hybrid intelligent systems. Of particular value to both the contributors and the readership are the short publication timeframe and the world-wide distribution, which enable both wide and rapid dissemination of research output. More information about this series at

3 László T. Kóczy Jesús Medina Editors Interactions Between Computational Intelligence and Mathematics 123

4 Editors László T. Kóczy Faculty of Engineering Sciences Széchenyi István University Gyõr Hungary and Jesús Medina Departamento de Matemáticas, Facultad de Ciencias Universidad de Cádiz Puerto Real, Cádiz Spain Budapest University of Technology and Economics Budapest Hungary ISSN X ISSN (electronic) Studies in Computational Intelligence ISBN ISBN (ebook) Library of Congress Control Number: Springer International Publishing AG 2018 This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, express or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Printed on acid-free paper This Springer imprint is published by Springer Nature The registered company is Springer International Publishing AG The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland

5 Preface Computational intelligence and traditional mathematics are two interwoven and extremely important scientific areas nowadays. For example, the treatment of big amount of data is an eminent goal in numerous topics which have been taken into consideration in important funding programs, such as the current European research and innovation program, Horizon 2020, which includes this problem in many of its subtopics. Hence, the interactions between computational intelligence and mathematics are very important in order to tackle a variety of important challenges. This volume is a new step in this direction including interesting contributions of engineers and mathematicians focused on the resolution of significant and actual computer science problems. The first paper deals with a social network-related problem: the investigation of the two most up-to-date ranking algorithms (PageRank and Katz) used for identifying the most influential participants in the discussion on a particular topic. The research is based on a case study, namely, the London Riots, using the Twitter database. In this study, most influential participating persons in the discussion are considered as trendsetters, i.e., having the ability to change the opinion of many others. The frequency of mentioning a participant s name and forwarding his/her tweets is used as the main measurable feature of a participant. After introducing the influence measures used by the two ranking algorithms mentioned above, the authors described their own experiments and results. They point out that for most users, the two methods agree but sometimes there are some surprising differences in the rank. While Katz uses simple frequency proportional mention weights, PageRank penalizes those mentions which come from less important participants. As the result of a deeper analysis, the authors conclude that PageRank has proved to be more robust a solution to identify influential users than the other top approaches. The second paper deals with the problem of the Synthetic Aperture Radar (SAR) imagery filtering. The problem in SAR images is the presence of atmospheric conditions introducing noise and partial covering by clouds. In addition, a large amount of speckle, a multiplicative non-gaussian noise degrades the quality of the SAR images. The study compares two classes of aggregation operators: the v

6 vi Preface Weighted Mean (WM) and the Weighted Ordered Average (OWA). A new family, merging the advantages of both previously mentioned classes, the Weighted OWA Operators (WOWA), offers a combination of advantages of the abovementioned two methods. The paper describes in detail the procedure of filtering the SAR images, with all three approaches. For learning the parameters, the authors use a Genetic Algorithm (GA), the classical evolutionary approach. As the GA is not one of the most efficient evolutionary methods, it is no surprise that the authors conclude that there is a strong dependence of the quality of the results on the number of generations. The final result points toward all proposed filters being useful, or even at the best depending on the type of image and the circumstances of application. The directions of future research are indicated at the end. The third paper proposes a novel combination of wavelet transform and fuzzy rule interpolation for evaluating very big data obtained from the telecommunication field. The engineering problem is the pre-evaluation of copper wire pairs from the point of view of transmission speed in SHDSL transmission networks. In the present best practice, the insertion loss must be measured at a large number of different frequencies for every wire pair. Based on several 100,000s of real measurement results, it is proposed that the combination of the Haar and Daubechies-4 wavelets combined with the Stabilized KH rule interpolation leads to a much better prediction of the transmission quality than any method known so far in the literature. The conclusion suggests that this new method might be useful in a much wider field of big data analysis, as well. The fourth paper introduces a logic called the Molecular Interaction Logic, which semantically characterizes the Molecular Interaction Maps (MIM) and, moreover, makes it possible to apply deductive and abductive reasoning on MIMs in order to find inconsistencies, answer queries, and infer important properties about those networks. This logic can be applied to different metabolic networks, as the one given by the cancer, which can sometimes appear in a cell as a result of some pathology in a metabolic pathway. The fifth paper is based on another logic. In this case in the general fuzzy logic programming called multi-adjoint logic programming. This framework was introduced by Medina et al. in 2001 as a general framework in which the minimal mathematical requirements are only considered in order to ensure the main properties given in the diverse usual logic programming frameworks, such as in possibilistic logic programming, monotonic and residuated logic programming, fuzzy logic programming, etc. Since its introduction, this framework has been widely developed from the theoretical and applied aspects by diverse authors. In this paper, the authors perform experiments which have shown the benefits of using a new c-unfolding transformation which reuses some variants of program transformation techniques based on unfolding, which have been largely exploited in the pure functional not fuzzy setting. Specifically, this new c-unfolding transformation has been applied to fuzzy connectives, and the authors have shown how to improve the efficiency of the proper unfolding process by reusing the very well-known concept of dependency graph. Furthermore, the paper includes a cost analysis and discussions on practical aspects.

7 Preface vii The last two papers are in the area of the Formal Concept Analysis (FCA). This framework arose as a mathematical theory for qualitative data analysis and has become an interesting research topic both on its mathematical foundations and on its multiple applications. Specifically, the sixth paper provides an overview of different generalizations of formal concept analysis based on fuzzy sets. First of all, a common platform for early fuzzy approaches is included. Then, different recently fuzzy extensions have been recalled, such as the generalized extension given by Krajči, the multi-adjoint concept lattices introduced by Medina et al., heterogeneous extension studied by diverse research groups exemplified by those of Krajči, Medina et al., Pócs, Popescu, using heterogeneous one-sided extension given by Butka and Pócs, and the higher order extension presented by Krídlo et al. The paper also discusses connections between the related approaches. The seventh paper is related to knowledge extraction from databases using rules, which is a compact tool in the representation of the knowledge. Based on concepts from lattice theory, this paper has introduced a new kind of attribute implications considering the fuzzy notions of support and confidence. The authors have also studied different properties of the particular case in which the set of attributes are intensions and, finally, they have also included an application to clustering for size reduction of concept lattices. We would like to express our gratitude to all the authors for their interesting, novel, and inspiring contributions. Peer-reviewers also deserve our deep appreciation because their deep and valuable remarks and suggestions have considerably improved the contributions. And last but not least, we wish to thank Dr. Tom Ditzinger, Dr. Leontina di Cecco, and Mr. Holger Schaepe for their dedication and help to implement and finish this large publication project on time maintaining the highest publication standards. Gyõr, Budapest, Hungary Puerto Real, Cádiz, Spain László T. Kóczy Jesús Medina

8 Contents Page Rank Versus Katz: Is the Centrality Algorithm Choice Relevant to Measure User Influence in Twitter?... 1 Hugo Rosa, Joao P. Carvalho, Ramon Astudillo and Fernando Batista Weighted Means Based Filters for SAR Imagery L. Torres, J. C. Becceneri, C. C. Freitas, S. J. S. Sant Anna and S. Sandri On Combination of Wavelet Transformation and Stabilized KH Interpolation for Fuzzy Inferences Based on High Dimensional Sampled Functions Ferenc Lilik, Szilvia Nagy and László T. Kóczy Abductive Reasoning on Molecular Interaction Maps Jean-Marc Alliot, Robert Demolombe, Luis Fariñas del Cerro, Martín Diéguez and Naji Obeid Efficient Unfolding of Fuzzy Connectives for Multi-adjoint Logic Programs Pedro J. Morcillo and Ginés Moreno On Fuzzy Generalizations of Concept Lattices Lubomir Antoni, Stanislav Krajči and Ondrej Krídlo Generating Fuzzy Attribute Rules Via Fuzzy Formal Concept Analysis ValentínLiñeiro-Barea, Jesús Medina and Inmaculada Medina-Bulo ix