Visit and STRC seminar not to miss next Tuesday: Image and Signal Processing Group of University of Valencia (26th June at 15h 30’ in PG16)

Dear Colleagues,
I would like to draw your attention to the visit of two members and a postdoc researcher of the Image and Signal Processing Group of University of Valencia, Spain.
This group is very influential in several areas like for instance Image Processing (in Geosciences, Medical Brain Imaging, etc.) and Kernel Machines; and they will be in Talbot Campus just for one day.
They will deliver a couple of short talks and after that they will stay for an hour for discussing ideas or future plans with anybody interested in BU.
The special seminar will take place next Tuesday, the 26th of June in PG 16 Lecture Theatre at 15:30 h (Ground floor, Poole House)
After that, you are very welcome to join us in an informal Discussion Panel from 16h 30’ to 17h 15’ approx.
I would like to encourage DEC PhD researchers, senior research fellows and staff to attend; particularly those who work or what to get into image/signal processing and kernel machines because they are leading experts in these areas. Kindly check out, for instance,
The agenda of the visit is the following:
• 15h 30’. Short intro by Dr. Malo (Associate Prof): “Research at the Image and Signal Processing Group”. Jesús Malo. A brief overview of our research interests and lines.
• 15h 40’. Short talk by Dr Laparra (Postdoc): “Gaussianization Framework for Signal Processing”
Abstract: We generalize a class of projection pursuit methods to transform arbitrary multidimensional data into multivariate normal data, thus attaining statistical independence of its components. The proposed analysis enables a number of novel ways to solve practical problems in high-dimensional scenarios, such as those encountered in image processing, speech recognition, array processing, or bioinformatics. Our framework extends Independent and Principal Components Analyses-based methods, which are typically not applicable to data generated from nonlinear, non-independent or non-Gaussian sources. The performance is successfully illustrated in a number of multidimensional data processing problems such as image synthesis, classification, saliency analysis, and de-noising.
• 16h. Short talk by Dr. Camps (Associate Prof): “Extended Kernel Methods”.
Abstract: I will talk about our love story with kernel methods for the last 10 years. Kernel methods constitute a simple way of translating linear algorithms into nonlinear ones. I will revise several interesting developments for 1) time series analysis, regression and function approximation; 2) classification problems; 3) nonlinear feature extraction; and 4) dependence estimation. The introduced methods extend previous standard algorithms to deal with non-stationary environments and structured domains, and the presence of non-Gaussian noise. Additionally, I’ll briefly talk about a way to learn the kernel function directly from the data via clustering or graphs. Examples in signal and image processing will guide this overview.
• 16h 25’. Discussion Panel.
Please, feel free to show up or leave any time during this event on your convenience. I hope you consider this program attractive and that you find a slot to come in.
Best Wishes, Emili