Following the success of Deep Learning (DL) in other domains, efforts to incorporate DL techniques to physical layer communication, in particular, are being carried out recently. In this regard, the research interests can roughly be categorized into two, i.e. End-to-End Design approach and Augmenting the existing algorithms with DL. Our lab aims to study how DL techniques can be incorporated to system design and optimization for 5G & Beyond 5G technologies. Motivated by the DARPA Spectrum Collaboration Challenge(SC2), which is a collaborative machine-learning competition, our lab aims to investigate a new and more efficient wireless paradigm in which radio networks automatically collaborate to dynamically determine how spectrum should be used in adaptive manner, instead of the current rigid and exclusively licensed bands.