Algorithms for optimization and machine learning over cloud

Abstract

Considering avenues and application areas of automation there is a significant push towards research advance in data mining and machine learning areas. Many relevant problems in these areas are computationally hard - generally modeled as optimization problems. These problems can be solved using exact algorithms or with the help of meta-heuristics - designed and inspired by natural computing. In this chapter, we consider two optimization problems - considered to be most central to machine learning and data mining algorithms design - the PCA and LDA computation. We discuss solving these problems in exact way over the cloud environment. We also present these computations when the data is available in streaming fashion. Finally, we present related probabilistic generative models (PPCA and PLDA) and show comparative study between these algorithms with implemented experiments and related results.