The Learning and Inference Network Architecture for 6G Wireless
In this talk, we present the deep learning based AI technologies and associated open issues related with the 6G wireless. The distributed learning and Inference computing and the networking aspects will be one of the driving factor to design 6G networks. The theoretical foundations for the network-based distributed AI will be discussed, such as information-funnel and information bottle-necks, the training data privacy and inferencing trustworthiness. We propose a 6G networking architecture to natively support the System-1 AI and can be evolved into System-2 AI.