Książka z serii "Rozprawy Doktorskie. Monografie"
Biclustering is an advanced clustering technique that can be applied to data mining and pattern recognition. Biclustering algorithms analyse rows and columns of an input matrix simultaneously, hence the ‘bi’ prefix. This is in contrast with other clustering methods that cluster rows and columns separately. Uniquely, biclustering produces local patterns, i.e. subsets of rows and subsets of columns. This ability makes biclustering techniques particularly promising for applications in bioinformatics and personalized medicine, where subgroup identification is a particular challenge.
In this monograph, the foundations of biclustering are reviewed and novel ‘hybrid’ biclustering algorithms are introduced. These proposed ‘hybrid’ methodologies combine key elements from other successful approaches. This yields a more versatile algorithm allowing application to multiple fields including socialnetworks, or business process optimization. This work describes why hybrid biclustering algorithms are important, explains the design of these methods, and provides examples of their application.