Title

Machine learning and statistical methods for multimedia QoE measurement and enhancement

Organizers

Tiago H. Falk, Zhi Li, and Lucjan Janowski

Motivation and objectives

It is without a doubt that data‐driven machine learning methods, in particular deep neural networks, have revolutionized numerous multimedia applications, including speech recognition and image/video-based object recognition, to name a few. Recently, efforts have emerged that explore the use deep neural networks for multimedia enhancement, blind source separation, room acoustics characterization, context‐awareness, and objective quality and quality‐of‐experience prediction. Recent challenges, however, suggest that there is still ample room for improvement, particularly within the domain of multimedia enhancement and quality assessment, as DNN‐enhanced signals have unusual distortions that existing metrics are not able to model well.

Topics of interest

Topics of Interest (not limited to):

Machine‐learning based multimedia signal enhancement (e.g., audio, video, image)
Machine‐learning based source separation
Data‐driven objective multimedia quality metrics
Machine learning and statistical methods for QoE modeling
Statistical methods for quality measurement and prediction
Data‐driven influential factor modeling for QoE
Characterizing distortions in DNN‐based methods
Machine learning and statistical methods for subjective QoE assessment