Delete search term

Header

Quick navigation

Main navigation

Dr. Siavash Arjomand Bigdeli: Deep Density Approximation for Bayesian Image Restoration, 09/12/2019 13:00

At Thursday, 12th September from 13:00 to 14:30, we have a very interesting talk from Dr. Siavash Arjomand Bigdeli (CSEM)- The event takes place in the room TE 402.

Title: Deep Density Approximation for Bayesian Image Restoration

Abstract: Finding strong oracle priors is an important topic in image restoration. In this talk, I will show how denoising autoencoders (DAEs) learn to mean-shift in O(1), and how we leverage this to employ DAEs as generic priors for image restoration. I will also discuss the case of Gaussian DAEs in a Bayesian framework, where the degradation noise and/or blur kernel are unknown. Experimental results demonstrate state of the art performance of the proposed DAE priors.

About the speaker: Siavash Bigdeli received his Ph.D. in Computer Science in 2018 from the Bern University. Following his postdoctoral fellowship at Ecole Polytechnique Fédérale de Lausanne (EPFL), he joined csem in 2019 as a data scientist in the Embedded Vision Systems group. He works on production-ready signal processing algorithms, with the focus on unsupervised deep learning solutions.

For more Information see Siavashs Homepage https://ivrlwww.epfl.ch/arjomand/