A structured view on pattern mining-based biclustering

Mining matrices to find relevant biclusters, subsets of rows exhibiting a coherent pattern over a subset of columns, is a critical task for a wide-set of biomedical and social applications. Since biclustering is a challenging combinatorial optimization task, existing approaches place restrictions on the allowed structure, coherence and quality of biclusters. Biclustering approaches relying on pattern mining (PM) allow an exhaustive yet efficient space exploration together with the possibility to discover flexible structures of biclusters with parameterizable coherency and noise-tolerance. Still, state-of-the-art contributions are dispersed and the potential of their integration remains unclear. This work proposes a structured and integrated view of the contributions of state-of-the-art PM-based biclustering approaches, makes available a set of principles for a guided definition of new PM-based biclustering approaches, and discusses their relevance for applications in pattern recognition. Empirical evidence shows that these principles guarantee the robustness, efficiency and flexibility of PM-based biclustering. (C) 2015 Elsevier Ltd. All rights reserved.