conditional distribution learning with neural networks and its application to universal denoising

Department: Electrical & Computer Engineering
Faculty Advisor(s): Young-Han Kim

Primary Student
Name: Jongha Ryu
Phone: 858-291-2433
Grad Year: 2021

A simple and scalable denoising algorithm is proposed that can be applied to a wide range of source and noise models. At the core of the proposed CUDE algorithm is symbol-by-symbol universal denoising used by the celebrated DUDE algorithm, whereby the optimal estimate of the source from an unknown distribution is computed by inverting the empirical distribution of the noisy observation sequence by a deep neural network, which naturally and implicitly aggregates multiple contexts of similar characteristics and estimates the conditional distribution more accurately. Universality of the CUDE algorithm is also shown, based on that the CUDE algorithm can attain the performance of the DUDE algorithm if a neural network has a large enough capacity. The performance of CUDE is evaluated for grayscale images of varying bit depths, which improves upon DUDE and its recent neural network based extension, Neural DUDE.

Industry Application Area(s)
Control Systems | Software, Analytics

« Back to Posters or Search Results