Abstract: Machine learning (ML) is envisaged to be a key enabler for 6G. To cater for high ML inference accuracy at scale under limited local observations, it is essential to communicate local knowledge over ML agents, where the knowledge lies within ML models as well as the data flows propagating through the models. This underscores the importance of developing communiication-efficient and energy-efficient distributed ML by understanding ML model architectures. In classical ML, the model is a neural network (NN), and the data flows are hidden layer activations, which can be communicated by federated learning (FL) and split learning (SL) frameworks, respectively. In the looming quantum ML, the model is a parameterized quantum cirquit (PQC), also known as a quantum NN (QNN), and the flows are qubits. Towards taming this new kind of data traffic, this talk aims to provide a brief overview of FL and SL, and present their communication-efficient applications under the state-of-the-art ML architectures such as visual transformer (ViT) and slimmable quantum NN (SQNN), as well as the centralized training with decentraliezd execution (CTDE) architecture for multi-agent reinforcement learning.

Bio: Jihong Park is a Lecturer at the School of IT, Deakin University, Australia. He received the B.S. and Ph.D. degrees from Yonsei University, Seoul, Korea, in 2009 and 2016, respectively. He was a Post-Doctoral Researcher with Aalborg University, Denmark, from 2016 to 2017; the University of Oulu, Finland, from 2018 to 2019. His recent research focus includes distributed machine learning, control, and resource management, as well as their applications to 6G semantic, AI-native, and non-terrestrial communications. He served as a Conference/Workshop Program Committee Member for IEEE GLOBECOM, ICC, and INFOCOM, as well as NeurIPS, ICML, and IJCAI. He received the IEEE GLOBECOM Student Travel Grant and the IEEE Seoul Section Student Paper Contest Bronze Prize in 2014, the 6th IDIS-ETNEWS Paper Award, and FL-IJCAI Best Student Paper Award in 2022. Currently, he is an Associate Editor of Frontiers in Data Science for Communications and in Signal Processing for Communications. He is a Senior Member of IEEE and a Member of ACM.