Abstract | ||
---|---|---|
Reconstructing the interacting structure of complex networks from available data is fundamental to understanding and controlling its collective dynamics. From the perspective of computational complexity, most network reconstruction problems are non-convex, making them difficult to be efficiently solved. The majority of existing approaches extend this problem to the convex optimization problem, which leads to a final solution that is far from being completely and precisely consummated. To improve the accuracy of network reconstruction, particularly when existing algorithms cannot fully reconstruct networks, a memetic algorithm (MA) is first proposed to solve this non-convex problem directly, termed as MAST-Net. According to the problem characteristics, correction and local search operators are designed to accelerate the MA convergence speed. We apply MAST-Net to evolutionary game models, resistor networks, and communication networks taking place in synthetic and real networks and demonstrate that MAST-Net exhibits competitive performance against seven state-of-the-art methods in terms of effectiveness and efficiency. |
Year | DOI | Venue |
---|---|---|
2019 | 10.1016/j.knosys.2018.11.009 | Knowledge-Based Systems |
Keywords | Field | DocType |
Network reconstruction,Memetic algorithm,Non-convex optimization | Convergence (routing),Memetic algorithm,Data mining,Telecommunications network,Computer science,Theoretical computer science,Complex network,Operator (computer programming),Local search (optimization),Convex optimization,Computational complexity theory | Journal |
Volume | ISSN | Citations |
164 | 0950-7051 | 3 |
PageRank | References | Authors |
0.36 | 14 | 3 |