Séminaire I&M, par Karim Seghouane (University of Melbourne, Australia)

Karim Seghouane

Titre : Learning Robust and Sparse Principal Components

ABSTRACT :

In this seminar, novel robust principal component analysis (RPCA) methods to exploit the local structure of datasets are presented. The proposed methods are derived by minimizing the α-divergence between the sample distribution and the Gaussian density model. The α−divergence is used in different frameworks to represent variants of RPCA approaches including orthogonal, non-orthogonal, and sparse methods. We show that the classical PCA is a special case of the proposed methods where the α−divergence is reduced to the Kullback-Leibler (KL) divergence. Furthermore, I shown in simulations that the proposed approaches recover the underlying principal components (PCs) by down-weighting the importance of structured and unstructured outliers. Using simulated data, it is shown that the proposed methods can be applied to fMRI signal recovery and Foreground-Background (FB) separation in video analysis. Results on real world problems of FB separation as well as image reconstruction are also presented.

BIOGRAPHIC:

Dr. Abd-Krim (Karim) Seghouane is a faculty member in the School of Mathematics and Statistics at the University of Melbourne, Australia.
Prior to this, he was a Senior Lecturer at the Department of Electrical and Electronic Engineering at the same University. He received his PhD from Paris Sud University, now known as University of Paris-Saclay, France. Upon completing his PhD, he worked as postdoctoral researcher at INRIA Rocquencourt, France and then as a researcher and subsequently as a senior researcher with National ICT Australia (NICTA). While at NICTA he was also an adjunct faculty member with the College of Engineering and Computer Science at the Australian National University (ANU). Dr. Seghouane’s research interests are in the areas of statistical signal and image processing, machine learning and artificial intelligence. His current research applications are focused on medical imaging and physiological signal analysis. He has received fellowships from the Australian Research Council, the Japanese Society for the Promotion of Science, the Australian Academy of Science and the French National Institute for Research in Digital Science and Technology.
Dr. Seghouane is currently an elected member of the IEEE Signal Processing Society Computational Imaging Technical Committee (CITC) and prior to that he was an elected member of the IEEE Signal Processing Society Machine Learning for Signal Processing Technical Committee (MLSPTC). While he was a member of the IEEE MLSPTC, he was also the chair of the data competition sub-committee. He was the General Co-Chair of the 2014 IEEE Workshop on Statistical Signal Processing and of the 2021 IEEE Workshop on Machine Learning for signal Processing, Gold Coast, Australia, the Data Competition Co-Chair of the 2017 and 2018 IEEE Workshop on Machine Learning for Signal Processing, Tokyo, Japan and Aalborg, Denmark, and the Organization Co-Chair of the 2017 IEEE International Symposium on Biomedical Imaging, Melbourne, Australia. He has served as an Associate Editor on the editorial board of the IEEE Transactions on Image Processing, from 2014 to 2018; and since 2018, he serves as a Senior Editor Area. He has also served on the editorial board of the IEEE Transactions on Signal Processing as an Associate Editor, from 2017 to 2021.