Continued research in the area of unsupervised and semi-supervised learning using kernel-based methods particularly in the context of spectral clustering. Developed novel models incorporating sparsity, prior knowledge and side information and hierarchical representations.
K
Postdoctoral Researcher
KULEUVEN
Jan 2009 - Oct 2010(1 year 10 months)
K
Ph.D student
KULEUVEN
Jan 2005 - Dec 2009(5 years)
Conducted research in the area of kernel methods and unsupervised learning leading to doctoral thesis: "Support Vector Methods for Unsupervised Learning". Main results can be summarized as nonlinear primal - dual models for several unsupervised learning problems such as feature extraction, dimensionality reduction, denoising and clustering in the style of support vector machines. The novelty corresponds to the incorporation of advanced concepts such as out-of-sample extensions and model selection into the models.