Human-Centered Multimedia Computing

The amount of multimedia data (including images, videos, music, speech, etc.) has been exploding around us. Since multimedia data are ultimately consumed by human users, how to process and manage such data in the viewpoint of humans is an important research issue. Toward human-centered multimedia services, we work on developing multimedia processing technologies based on human perceptual mechanisms, including perceptual quality evaluation, image/video analysis, content recommendation, image/video enhancement, mobile video transmission, etc.

Human-Like Machine Learning

Implementation of multimedia processing techniques often requires human-like computing. In particular, we conduct research on machine learning, which mimics humans' capability of learning generalized knowledge from examples. Topics include neural networks, deep learning, multimodal data learning, bio-inspired stochastic optimization, etc.