Discrete bacterial foraging optimization for community detection in networks

Abstract

An essential mesoscopic concept in network analysis is that of community structure. However, conventional nature-inspired optimization algorithms encounter serious challenges and difficulties when used directly to seek communities in networks, due to the large amount of data and the NP-hard combinatorial nature of the problem. Thus in this paper, we introduce a novel bacterial foraging optimization approach to uncovering community structure in networks. Instead of using the original bacterial foraging designed traditionally for continuous optimization, the problem of community detection is exquisitely embedded into a redefined discrete framework. The evolutionary principles for the bacterial foraging are developed from a topological perspective. Furthermore, two specific local updating rules, namely the greedy strategy and the stochastic strategy, are designed to steer the swarm of bacteria to the favored regions. The extensive experimental results on both synthetic and real-world networks indicate that the proposed approach outperforms the baseline algorithms and can achieve a high accuracy on the uncovered community structure. The integration of the proposed approach into the analysis of power grids and its explicit utility are also discussed in detail, showing that our method has high accuracy and practicability.