Consensus Clustering Paradigm for Brain Networks

Clustering algorithms can extract information from large datasets through model-free or data-driven approaches. However, in applications with real data with little a priori knowledge, it is often difficult to select an appropriate clustering algorithm and evaluate the quality of clustering results due to the unknown ground truth. It is also the case that conclusions based on only one specific algorithm might be biased, since each algorithm has its own assumptions of the structure of the data, which might not correspond to the real data. In cases of multiple heterogeneous datasets from similar experiments, which may have been generated either in the same laboratory or different laboratories, the challenge is how to reach consensus conclusions. This presentation will address these issues and report