Advanced kernel-based modeling for unsupervised and semi-supervised learning.

This research proposal is aimed at addressing key challenges and open problems in applications of unsupervised and semi-supervised learning using advanced kernel-based models. This type of machine learning paradigm has gained recently a considerable amount of interest due mainly to its successful application to high-dimensional data and the possibility to include nonlinearities into the analysis in an efficient way. The research objectives that will be investigated are summarized in the following aspects: developing a unifying framework to link different formulations, incorporating prior knowledge and side information into the analysis, formulating sparser models to cope with the exponentially increasing amount of data available nowadays, developing and designing model selection criteria for a broader type of methods and applications and improving the current kernel models. We will use the optimization perspective as a core methodology to fulfill the aformentaioned objectives.

Keywords:Model selection, Optimization, Semi-supervised learning, Unsupervised learning, Kernel methods, Machine learning

Disciplines:Artificial intelligence, Cognitive science and intelligent systems, Applied mathematics in specific fields, Computer architecture and networks, Distributed computing, Information sciences, Information systems, Programming languages, Scientific computing, Theoretical computer science, Visual computing, Other information and computing sciences