Skip to content

Report from FCN 2020

The FCN2020 workshop on Edge Intelligence, IoT systems and Analytics

The workshop was held online on 17 August 2020. TECoSA PI György Dán took part, and brings us this summary.

The program included six keynote lectures intertwined with breakout sessions on methodological and systems aspects of edge computing, data analytics and IoT, and two panel discussions.

Gyorgy Dan
TECoSA PI, György Dán

Among the keynotes, Nic Lane from the University of Cambridge and Samsung AI talked about their framework and API for the deployment of federated learning applications in the cloud, and its use for the evaluation of the carbon footprint of edge based federated learning.

Ferran Diego from Telefonica provided an experimental look at the tradeoff between choosing machine learning models and execution platforms, as a function of their location and the latency requirements of the applications, highlighting the need for adaptive application aware placement.

Iqbal Mohomed from Samsung AI, Canada, talked about using home devices, such as mobile phones, for performing machine learning in home environments, the issues in discovering resources and in optimizing the use of distributed resources for optimizing system performance.

Roberto Morabito from Princeton University and Ericsson talked about the design of a framework for resource aware placement of inference tasks in edge computing environments, focusing on orchestration and performance profiling.

The panels accompanying the presentations highlighted issues concerning accountability and trustworthiness of the edge infrastructure, how to achieve scalability in terms of performance, the need for toolkits for accelerating application development and deployment, including standards and interoperability. Arguments in the panel for ensuring the success of edge computing included the need for ease of accessibility to programmers, unified APIs, the need for development resources, programming models and resource aggregators, topics that are well aligned with the objectives of TECoSA.

https://cpi-lab.github.io/fcn/2020/