Consensus Community Detection in Multilayer Networks Using Parameter-Free Graph Pruning

Abstract

The clustering ensemble paradigm has emerged as an effective tool for community detection in multilayer networks, which allows for producing consensus solutions that are designed to be more robust to the algorithmic selection and configuration bias. However, one limitation is related to the dependency on a co-association threshold that controls the degree of consensus in the community structure solution. The goal of this work is to overcome this limitation with a new framework of ensemblebased multilayer community detection, which features parameter-free identification of consensus communities based on generative models of graph pruning that are able to filter out noisy co-associations. We also present an enhanced version of the modularity-driven ensemble-based multilayer community detection method, in which community memberships of nodes are reconsidered to optimize the multilayer modularity of the consensus solution. Experimental evidence on real-world networks confirms the beneficial effect of using model-based filtering methods and also shows the superiority of the proposed method on state-of-the-art multilayer community detection.

Modularity in Multilayer Networks Using Redundancy-Based Resolution and Projection-Based Inter-Layer Coupling

“…Due to different size of our evaluation datasets, we devised several configurations of variation of parameter k in PMM, by reasonably adapting each of the configuration range and increment step to the network size. Concerning M-EMCD * , we used the marginal likelihood filter (MLF) to perform parameter-free detection of the number of communities [20].…”

“…As exemplary methods of the aggregation approach, we used Principal Modularity Maximization (PMM) [48] and Enhanced Modularity-driven Ensemblebased Multilayer Community Detection (M-EMCD * ) [20]. PMM aims to find a concise representation of features from the various layers (dimensions) through structural feature extraction and cross-dimension integration.…”

“…It should be emphasized that neglecting such a kind of complex organization by reducing the whole system to a single network (e.g., through some kind of projection, or by aggregation), has been shown to be much less informative than the multilayer representation [14]. For the above mentioned reasons, multilayer networks are experiencing an increasing interest from the scientific community, leading to an explosion of scientific papers in many areas of science, thus becoming one of the most used tools for interdisciplinary research [15], [16], [11], [13], [17], [18], [12], [19], [20,21], [22], [23].…”

Multilayer network simplification: Approaches, models and methods

“…It can also account for known factors that influence the occurrence of interactions, such as known group structures, similarities between elements, or other forms of biases. Some of the above generative models for graph pruning have been recently used in the context of consensus community detection in multilayer networks [80]. Essentially, a generative null model is evaluated on a weighted graph of coassociations (or co-occurrences): given an input multilayer network and an ensemble of layer-specific community structures defined over it, a weighted coassociation graph is an undirected graph whose nodes correspond to the actors/entities in the multilayer network and the strength of an edge corresponds to the fraction of communities that two entities share in the ensemble community structures.…”

DAOC: Stable Clustering of Large Networks

“…a) Robust clustering algorithms: typically leverage consensus or ensemble techniques [6]- [9]. They identify clusters using consensus functions (e.g., majority voting) by processing an input network multiple times and varying either the parameters of the algorithm, or the clustering algorithm itself.…”