Fast machine learning algorithms for large data

摘要: Traditional machine learning has been largely concerned with developing techniques for small or modestly sized datasets. These fail to scale up well large data problems, a situation becoming increasingly common in today’s world. This thesis is the problem of data. In particular, it considers solving three basic tasks learning, viz., classification, regression and density approximation. We develop fast memory- efficient algorithmics kernel training deployment. include considering preprocessing steps speeding existing algorithms as general purpose framework using methods. Emphasis placed on development computationally greedy schemes which leverage state-of-the-art from field numerical linear algebra. The presented here underline premise that possible efficiently train data, generalizes yet sparse expansion leading improved runtime performance. Empirical evidence provided support this throughout thesis.