Classifiers of massive and structured data problems: algorithms and applications

The last two decades have seen the emergence of vast and unprecedented data repositories. Extraordinary opportunities now present themselves for new data analysis methods that can harness these repositories. As larger and larger amounts of widely varying types of data are constantly being collected and assimilated, the task of making use of such data opens up interesting and challenging avenues of research. This thesis deals with specific problems in data mining and machine learning in this setting. In particular we describe algorithms and applications for classification problems where computational restrictions become limiting (resource bounded algorithms and online/streaming algorithms) as well as models and algorithms for certain problems where the structure of the input is leveraged to provide not only accurate, but also interpretable classifiers.Ph.D.Includes bibliographical references