基于群体智能优化的大数据复杂网络结构分析

声明

摘要

ABSTRACT

List of Figures

List of Tables

Notations

Abbreviations

Contents

Chapter 1 Introduction

1．1 Background

1．2 Network Issues，Notations and Properties

1．2．1 Issues Concerning Network Analytics

1．2．2 Graph Based Network Notation

1．2．3 Eminent Properties of Network

1．3 Community Structure Analytics

1．3．1 Description of Community Discovery

1．3．2 Qualitative Community Definition

1．3．3 Existing Approaches for Community Discovery

1．4 Structural Balance Analytics

1．4．1 Signed Network Notation

1．4．2 Structural Balance Theory

1．4．3 The Importance of Structural Balance

1．5 Optimization and Evolutionary Algorithm

1．5．1 What Is Optimization

1．5．2 Why We Need Optimization

1．5．3 How to tackle Optimization Problems

1．5．4 Evolutionary Multiobjective Optimization

1．6 Particle Swarm Optimization

1．6．1 Canonical Particle Swarm Optimization

1．6．2 Discrete Particle Swarm Optimization

1．7 Multiobjective Particle Swarm Optimization

1．8 Organization of the Dissertation

Chapter 2 Unsigned Big Network Community Discovery Based on Particle Swarm Optimization

2．1 Motivation

2．2 Proposed Algorithm for Community Discovery

2．2．1 Algorithm Framework

2．2．2 Fitness Function

2．2．3 Particle Representation and Initialization

2．2．4 Particle-status-updating Rules

2．2．5 Particle Position Reordering

2．3 Experimental Study

2．3．1 Performance Metric

2．3．2 Results on Synthetic Networks

2．3．3 Results on Real-world Networks

2．4 Additional Discussion on GDPSO

2．4．1 Discussion on Algorithm Parameters

2．4．2 Discussion on Position Update Principle

2．5 Conclusions

Chapter 3 Signed Big Network Community Detection Based on Particle Swarm Optimization

3．1 Motivation

3．2 Proposed Algorithm for Community Discovery

3．3 Experimental Studies

3．3．1 Comparison Algorithms

3．3．2 Results on Synthetic Signed Networks

3．3．3 Results on Real-World Signed Networks

3．4 Conclusions

Chapter 4 Multi-Resolution Network Clustering Using MOPSO With Decomposition

4．1 Motivations

4．1．1 Motivations for Choosing PSO Framework for ComplexNetwork Clustering

4．1．2 Motivations for Proposing the Discrete MOPSO Algorithm

4．1．3 Motivations for Introduced Mechanisms to Preserve Diversity

4．2 Proposed Algorithm for Multi-Resolution Network Clustering

4．2．1 Objective Function

4．2．2 Definition of Discrete Position and Velocity

4．2．3 Discrete Particle Status Updating

4．2．4 Particle Swarm Initialization

4．2．5 Selection of Leaders

4．2．6 Framework of the Proposed Algorithm

4．2．7 Turbulence Operator

4．2．8 Complexity Analysis

4．3 Experimental Studies

4．3．1 Comparison Algorithms

4．3．2 Experimental Settings

4．3．3 Experiments on Unsigned Benchmark Networks

4．3．4 Experiments on Unsigned LFR Benchmark Networks

4．3．5 Experiments on Unsigned Real-world Networks

4．3．6 Experiments on Signed Networks

4．4 Conclusions

Chapter 5 A Two-Step Approach for Network Structural Balance Analytics

5．1 Motivation

5．1．1 Limitations of Traditional Methods

5．1．2 Our Two-Step Idea

5．2 Methodology

5．2．1 General Framework

5．2．2 Model Selection

5．2．3 Complexity Analysis

5．3 Experimental Study

5．3．1 Signed Network Data Sets

5．3．2 Validation Experiments

5．3．3 Comparisons With Other MOEAs

5．3．4 Structural Balance Experiments

5．3．5 Discussion on Parameters

5．4 Conclusions

Chapter 6 Conclusion And Perspectives

6．1 Thesis Conclusion

6．2 Future Directions and Challenges

References

Acknowledgments

Biography