Computational Intelligence for Network Structure Analytics

Computational Intelligence for Network Structure Analytics

Maoguo Gong Qing Cai Lijia Ma Shanfeng Wang Yu Lei Computational Intelligence for Network Structure Analytics 123

Maoguo Gong Xidian University Xi an, Shaanxi China Qing Cai Hong Kong Baptist University Hong Kong China Shanfeng Wang Xidian University Xi an, Shaanxi China Yu Lei Northwestern Polytechnical University Xi an, Shaanxi China Lijia Ma Hong Kong Baptist University Hong Kong China ISBN 978-981-10-4557-8 ISBN 978-981-10-4558-5 (ebook) DOI 10.1007/978-981-10-4558-5 Library of Congress Control Number: 2017950028 Springer Nature Singapore Pte Ltd. 2017 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 Nature Singapore Pte Ltd. The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721, Singapore

Preface Many real-world problems are actually optimization problems and it is our long standing quest to address these real-world optimization problems by means of techniques from diverse fields. Theories and techniques for canonical mathematical optimization have been well studied for centuries and play a leading role in solving optimization problems. Nevertheless, most if not all network issues are essentially hard optimization problems which very often cannot be well solved by traditional mathematical optimization techniques. The performances of canonical methods deteriorate rapidly especially when the real problems involve many optimization objectives and the number of decision variables is large. In order to remedy the drawbacks of canonical optimization techniques, computational intelligence, a class of artificial intelligence techniques, has come into being and is widely recognized as a promising computing paradigm. Evolutionary computation, an important branch of computational intelligence, emerges and tremendous efforts have been done to design many kinds of efficient evolutionary algorithms for solving diverse hard optimization problems. Apart from evolutionary computation, other computational intelligence techniques such as swarm intelligence, meta-heuristics, and artificial neural networks have all find their niche in the optimization field. This book makes efforts to delineate in detail the existing state-of-the-art methods based on computational intelligence for addressing issues related to complex network structures. Using computational intelligence techniques to address network issues may facilitate smart decisions making by providing multiple options to choose from, while conventional methods can only offer a decision maker a single suggestion. Meanwhile, evolutionary computation provides a promising outlet toward network issues and in turn network structure patterns may provide novel inspiration toward the design of next-generation computational intelligence techniques. As a comprehensive text, the contents of the whole book cover most emerging topics of both network structures analytics and evolutionary computation, including theories, models, algorithm design, and experimental exhibitions. This book summarizes the researches achievements on the topics by the authors, their postgraduate v

vi Preface students and their alumni ever since 2008. Offering a rich blend of theories and practices, the book is suitable for students, researchers, and practitioners interested in network analytics and computational intelligence, both as a textbook and as a reference work. We would like to take this great opportunity to extend our sincere thanks to the editors at Springer Press and the anonymous reviewers for their helpful comments for improving the quality of this book. Finally, we would like to thank all the members with the Collaborative Innovation Center for Computational Intelligence (OMEGA) at Xidian University. This research was supported by the National Natural Science Foundation of China (Grant nos. 61273317, 61422209, and 61473215), the National Program for Support of Top-notch Young Professionals of China, and the National key research and development program of China (Grant no. 2017YFB0802200). Xi an, China Hong Kong, China Hong Kong, China Xi an, China Xi an, China Maoguo Gong Qing Cai Lijia Ma Shanfeng Wang Yu Lei

Contents 1 Introduction.... 1 1.1 Network Structure Analytics with Computational Intelligence... 1 1.1.1 Concepts of Networks... 2 1.1.2 Community Structure and Its Detection in Complex Networks... 5 1.1.3 Structure Balance and Its Transformation in Complex Networks... 9 1.1.4 Network Robustness and Its Optimization in Complex Networks... 12 1.2 Book Structure... 14 References.... 15 2 Network Community Discovery with Evolutionary Single-Objective Optimization... 21 2.1 Review of the State of the Art... 21 2.2 A Node Learning-Based Memetic Algorithm for Community Discovery in Small-Scale Networks... 22 2.2.1 Memetic Algorithm with Node Learning for Community Discovery... 23 2.2.2 Problem Formation... 24 2.2.3 Representation and Initialization... 24 2.2.4 Genetic Operators... 25 2.2.5 The Local Search Procedure... 26 2.2.6 Experimental Results... 27 2.2.7 Conclusions.... 34 2.3 A Multilevel Learning-Based Memetic Algorithm for Community Discovery in Large-Scale Networks... 34 2.3.1 Memetic Algorithm with Multi-level Learning for Community Discovery.... 35 2.3.2 Representation and Initialization... 35 vii

viii Contents 2.3.3 Genetic Operators... 35 2.3.4 Multi-level Learning Strategies... 37 2.3.5 Complexity Analysis of MLCD.... 43 2.3.6 Comparisons Between MLCD and Meme-Net.... 44 2.3.7 Experimental Results... 45 2.3.8 Conclusions.... 52 2.4 A Swarm Learning-Based Optimization Algorithm for Community Discovery in Large-Scale Networks... 54 2.4.1 Greedy Particle Swarm Optimization for Network Community Discovery.... 55 2.4.2 Particle Representation and Initialization... 55 2.4.3 Particle-Status-Updating Rules.... 56 2.4.4 Particle Position Reordering... 58 2.4.5 Experimental Results... 59 2.4.6 Additional Discussion on GDPSO.... 65 2.4.7 Conclusions.... 71 References.... 71 3 Network Community Discovery with Evolutionary Multi-objective Optimization... 73 3.1 Review on the State of the Art... 73 3.2 A Decomposition Based Multi-objective Evolutionary Algorithm for Multi-resolution Community Discovery... 74 3.2.1 Multi-objective Evolutionary Algorithm for Community Discovery... 75 3.2.2 Problem Formation... 76 3.2.3 Representation and Initialization... 77 3.2.4 Genetic Operators... 78 3.2.5 Experimental Results... 78 3.2.6 Conclusions.... 84 3.3 A Multi-objective Immune Algorithm for Multi-resolution Community Discovery... 84 3.3.1 Multi-objective Immune Optimization for Multi-resolution Communities Identification... 85 3.3.2 Problem Formation... 85 3.3.3 Proportional Cloning... 86 3.3.4 Analysis of Computational Complexity... 88 3.3.5 Experimental Results... 88 3.3.6 Conclusions.... 96 3.4 An Efficient Multi-objective Discrete Particle Swarm Optimization for Multi-resolution Community Discovery... 97 3.4.1 Multi-objective Discrete Particle Swarm Optimization for Multi-resolution Community Discovery... 97 3.4.2 Problem Formation... 98

Contents ix 3.4.3 Definition of Discrete Position and Velocity... 99 3.4.4 Discrete Particle Status Updating.... 99 3.4.5 Particle Swarm Initialization... 102 3.4.6 Selection of Leaders... 102 3.4.7 Turbulence Operator... 103 3.4.8 Complexity Analysis... 103 3.4.9 Experimental Results... 104 3.4.10 Experimental Results on Signed Networks... 116 3.4.11 Conclusions.... 118 3.5 A Multi-objective Evolutionary Algorithm for Community Discovery in Dynamic Networks... 119 3.5.1 Multi-objective Optimization for Community Discovery in Dynamic Networks.... 119 3.5.2 Problem Formation... 120 3.5.3 Proportional Cloning... 121 3.5.4 Genetic Operators... 122 3.5.5 The Local Search Procedure... 122 3.5.6 Solution Selection... 124 3.5.7 Experimental Results... 125 3.5.8 Conclusions.... 131 References.... 133 4 Network Structure Balance Analytics with Evolutionary Optimization... 135 4.1 Review on the State of the Art... 135 4.2 Computing Global Structural Balance Based on Memetic Algorithm... 137 4.2.1 Memetic Algorithm for Computing Global Structural Balance... 137 4.2.2 Representation and Initialization... 138 4.2.3 Genetic Operators... 138 4.2.4 The Local Search Procedure... 139 4.2.5 Experimental Results... 141 4.2.6 Complexity Analysis... 145 4.2.7 Conclusions.... 146 4.3 Optimizing Dynamical Changes of Structural Balance Based on Memetic Algorithm.... 146 4.3.1 Problem Formation... 146 4.3.2 Representation and Initialization... 148 4.3.3 Genetic Operators... 149 4.3.4 The Local Search Procedure... 149 4.3.5 Transformation... 150 4.3.6 Experimental Results... 151 4.3.7 Conclusions.... 157

x Contents 4.4 Computing and Transforming Structural Balance Based on Memetic Algorithm... 158 4.4.1 Optimization Models... 159 4.4.2 Memetic Algorithm for the Computation and Transformation of Structural Balance in Signed Networks... 161 4.4.3 Experimental Results... 169 4.4.4 Conclusions.... 180 4.5 Computing and Transforming Structural Balance Based on Evolutionary Multi-objective Optimization... 181 4.5.1 The Two-Step Algorithm for Network Structural Balance... 182 4.5.2 Model Selection... 184 4.5.3 Complexity Analysis... 186 4.5.4 Experimental Results... 186 4.5.5 Conclusions.... 196 References.... 197 5 Network Robustness Analytics with Optimization... 201 5.1 Review on The State of the Art... 201 5.2 Enhancing Community Integrity Against Multilevel Targeted Attacks... 202 5.2.1 Model Malicious Attack on the Network as a Two-Level Targeted One.... 203 5.2.2 Community Robustness of Networks.... 205 5.2.3 Constraints for Improving Networks... 207 5.2.4 Enhancing Community Robustness of Networks... 208 5.2.5 Experimental Results... 209 5.2.6 Conclusions.... 217 5.3 Enhancing Robustness of Coupled Networks Under Targeted Recoveries... 217 5.3.1 Algorithm for Enhancing Robustness of Coupled Networks Under Targeted Recoveries... 218 5.3.2 Experimental Results... 223 5.3.3 Conclusions.... 226 References.... 227 6 Real-World Cases of Network Structure Analytics.... 229 6.1 Review on the State of the Art... 229 6.2 Community-Based Personalized Recommendation with Evolutionary Multiobjective Optimization... 232 6.2.1 MOEA-Based Recommendation Algorithm... 232 6.2.2 User Clustering... 232 6.2.3 Problem Formation... 233

Contents xi 6.2.4 Representation... 234 6.2.5 Genetic Operators... 234 6.2.6 Experimental Results... 235 6.2.7 Conclusions.... 244 6.3 Influence Maximization in Social Networks with Evolutionary Optimization... 244 6.3.1 Memetic Algorithm for Influence Maximization in Social Networks... 245 6.3.2 Network Clustering... 247 6.3.3 Candidate Selection... 247 6.3.4 Seed Generation... 248 6.3.5 Experimental Results... 253 6.3.6 Conclusions.... 259 6.4 Global Biological Network Alignment with Evolutionary Optimization... 259 6.4.1 Problem Formation... 260 6.4.2 Optimization Model for Biological Network Alignment... 260 6.4.3 Memetic Algorithm for Network Alignment... 261 6.4.4 Representation and Initialization... 262 6.4.5 Genetic Operators... 264 6.4.6 The Local Search Procedure... 264 6.4.7 Experiments Results... 266 6.4.8 Conclusions.... 277 References.... 277 7 Concluding Remarks.... 281 7.1 Future Directions and Challenges... 281