Compressed sensing based missing nodes prediction in temporal communication network
Abstract
The reconstruction of complex network topology is of great theoretical and practical significance. Most research so far focuses on the prediction of missing links. There are many mature algorithms for link prediction which have achieved good results, but research on the prediction of missing nodes has just begun. In this paper, we propose an algorithm for missing node prediction in complex networks. We detect the position of missing nodes based on their neighbor nodes under the theory of compressed sensing, and extend the algorithm to the case of multiple missing nodes using spectral clustering. Experiments on real public network datasets and simulated datasets show that our algorithm can detect the locations of hidden nodes effectively with high precision.
- Publication:
-
Physica A Statistical Mechanics and its Applications
- Pub Date:
- February 2018
- DOI:
- 10.1016/j.physa.2017.08.149
- Bibcode:
- 2018PhyA..492..265C
- Keywords:
-
- 00-01;
- 99-00