Temporal Block Spectral Clustering for Multi-Layer Temporal Functional Connectivity Networks

Many real world complex systems can be modeled as networks, i.e. graphs. A key approach to network analysis is community detection. Early work in community detection methods focused on a single network, whereas in most applications networks may be time dependent or may have multiple types of edges relating the nodes. Recently, multi-layer networks that incorporate multiple channels of connectivity have been introduced to represent such complex systems. In this paper, we focus on multi-layer temporal networks. A temporal block spectral clustering approach is proposed to detect and track the community structure across time. In this approach, both the connections between nodes of the network within a time window, i.e. intralayer adjacency, as well as the connections between nodes across different time windows, i.e. inter-layer adjacency are taken into account. The proposed framework is evaluated on both simulated and resting state fMRI data.