Information theoretic kernel machines for spectral clustering and deep learning research
We are pleased to present Robert Jenssen, from UiT Machine Learning Group who will speak at this month´s department seminar.
Main content
Abstract
This talk will present kernel machines, a dominant direction in machine learning, in terms of estimators of information theoretic quantities such as entropy and divergence. This will lead to the so-called kernel entropy component analysis method for spectral clustering. Moreover, the talk will present recent work for unsupervised training of deep discriminative neural networks based on divergence and regularization via kernels.
Biography
Robert Jenssen directs the Machine Learning @ UiT Lab. His research aim is to develop the next generation machine learning data analytics methodology, using information theoretic learning, kernel methods, graph spectral methods, and big data algorithms with deep learning. One the application side, he is especially focused on health data analytics by mining electronic patient journals.
In addition, he is
- Prof II at the Norwegian Computing Center (NR) in Oslo (please see NR News)
- Senior Researcher at the Norwegian Center on Integrated Care and Telemedicine, University Hospital of North Norway
