An Effective Resistance based Genetic Algorithm for Community Detection

This work presents a new approach based on genetic algorithms (GAs) and the concept of effective resistance for detecting communities within an undirected graph. The method considers the equivalent electric network of the input graph, where edges are weighted with their effective resistance, a measure of electrical resistance between nodes, whose square root has been shown to be a Euclidean metric. The algorithm computes the similarity between nodes by using the effective resistance values and generates a weighted and sparse graph by adopting a thresholding sparsification strategy based on the nearest neighbors of each node. Experiments over synthetic and real-world networks demonstrate the effectiveness of our approach when compared to other benchmark methods.