Abstract:
In this thesis we give a survey of online machine learning algorithms for data stream analysis. After giving an overview of standard batch algorithms, we explain batch-to-online conversion, and we give a in-depth description and analysis of data stream mining techniques. We particularly focus on online k-means algorithms and multilayer perceptron models as representative examples of online clustering and clas sification algorithms. We also present theoretical and empirical analyses of different approaches for online versions of these algorithms through numerical experiments.
