Privacy-Preserving Deep Learning
Deep learning is powered by abundant data. As it becomes more ubiquitous in smart cameras, CCTV, and home robots, deep learning is threatening people’s privacy by requiring people’s personal data, including their faces, names, locations, jobs, preferences, etc. Our goal is to fundamentally protect or preserve people’s privacy in collecting training/test data or in performing training/test/inference tasks. AiSLab has been collaborating with a start-up company EgoVid Inc., co-founded by Prof. Hyun Jong Yang, on this topic.
Drones will be pervasive around us in the close future. By virtue of their capability of changing location easily, they will be the key players in providing temporal or nomadic internet connectivity in rural areas, disaster scenes, or backwoods. Our goal is to develop drone base stations that can actively detect targets around them using radar and can serve wireless connectivity to users on the ground. We also develop intelligent systems such as autonomous Scanning Electron Microscope (SEM).
Signal Processing and Communication Theory
We study theory on optimization, communication, radar, computer vision, and machine learning. Theory is fundamental of real implementation. Our goal is to understand the state-of-the-art of novel algorithms or theoretical analysis in the fields of communication, radar, and computer vision, and to enhance them. We also prove our proposed concepts by developing test-bed systems.