PhD, École Polytechnique Fédérale de Lausanne

Applications of Matrix Factorization in signal processing

Matrix factorization-that is, representing a signal by suitable basis functions is widely used in machine learning, image denoising, signal processing and neuroscience. This problem usually comes with two main assumptions about the signal: the signal’s components either are independent or sparse. More precisely, in the independent-component analysis (ICA), the signal considered is a mixture of some independent sources. Source separation is one example for ICA. In contrast, in the sparse component analysis (SCA), the signal is assumed to be a mixture of sparse sources. We present an overview of the ICA and SCA from the matrix factorization point of view.

I received my B.Sc. in Electrical Engineering from Sharif University of Technology in 2014. Now, I am working towards my PhD in the Laboratory of Communications and Applications under the supervision of Prof. Thiran and Dr. Celis. My research interest revolves around Machine Learning. I am particularly interested in developing algorithms for fast Dictionary Learning and Multi-armed Bandits.

Watch video from here.