Machine learning techniques are increasingly being used with massive training data sets in application areas such as internet, retail, insurance, finance, manufacturing and life sciences. The Parallel Machine Learning (PML) toolkit is a software framework for machine learning algorithms on high-performance computer (HPC) platforms (such as the IBM Blue Gene/P supercomputer). Several well-known algorithms have been implemented using the PML framework to date, and we specifically describe the detailed implementation of the transform regression (TREG) algorithm, in view of its novelty, parallel scalability and wide applicability.
By: Sitaram Asur, Amol Ghoting, Ramesh Natarajan, Edwin Pednault
Published in: RC24882 in 2009
Questions about this service can be mailed to reports@us.ibm.com .
