On-Line Analytical Processing (OLAP) systems provide efficient low level database support for a variety of data analysis, machine learning and knowledge extraction tasks that involve very large datasets.

On-Line Analytical Processing (OLAP) systems provide efficient low level database support for a variety of data analysis, machine learning and knowledge extraction tasks that involve very large datasets. This talk will: (i) introduce the main concepts of OLAP, in particular the multi-dimensional data model that is relevant to machine learning, where dimensions correspond to ?features?; (ii) explain how the precomputation of various projections of the data can accelerate significantly the analysis of data; (iii) present the view selection problem that decides the set of projections to precompute in order to optimize performance under the constraints of storage and data maintenance time; and (iv) discuss how biologically inspired algorithms such as ant colony optimimization and Genetic algorithms, have been employed to solve the view selection problem.