CSE kollokvium: Recipes to Make Sustainable Machine Learning

The training and execution of Machine Learning (ML) models pose relevant challenges, as they often require substantial computational and energy resources. To mitigate the ML footprint on the computational resources of single network nodes, new distributed ML paradigms have emerged, as they can distribute the computational load across different network nodes, besides providing better privacy preservation. However, despite such advantages, distributed ML too may face significant hurdles, namely, (i) slower execution of the ML training or inference, and (ii) high bandwidth consumption on the links connecting the participating network nodes whenever the output of intermediate processing has to be transferred from one node contributing to the ML pipeline to another. In this talk, we discuss such issues and how they can be solved by leveraging dynamic neural networks.

Short CV

Carla Fabiana Chiasserini is Full Professor at Politecnico di Torino, Italy, and a Research Associate with the Italian National Research Council (CNR) and the National Inter-University Consortium for Telecommunications (CNIT). She is also a WASP Guest Professor at Chalmers Institute of Technology, Sweden. She was a Visiting Researcher at UC San Diego (1998-2003),and a Visiting Professor at Monash University (2012,2016),the Berlin Technische Universität (2021,2022),and the HPI, Potsdam University (Mercato Fellow,2023and 2024).She is an IEEE Fellow and an ACM Distinguished Member. She is the director of the Advanced Wireless Experience Lab at Politecnico di Torino. Her research interests include Next-generation Mobile Networks, Edge Computing, Machine Learning, and Mobile Autonomous Systems. She has published over 350 journal and conference papers and has received several awards for her scientific work. She serves as Vice Editor-in-Chief of Computer Communications and as Editor-at-Large of the IEEE/ACM Transactions on Networking.