1st Workshop on Distributed Edge AI – Risks and Challenges

(DE‑AI’23)

Warsaw, Poland, 17–20 September, 2023

For the last two decades the usual approach for IoT applications has been to leverage on cloud infrastructures to address the computational and storage limitations and constraints at end and edge nodes. However, offloading processing capabilities to the cloud requires transferring data from edge devices to a backend cloud infrastructure, dealing with the limitations imposed by the underlying communication channels, depending on the availability constraints of both those communication channels and the cloud infrastructure, establishing proper mechanisms to secure data in transit and data at rest at the backend, dealing with the costs incurred from the transmission of data and by the usage of the backend cloud infrastructure, etc.

The advent of end and edge nodes with increased computational and storage capabilities makes it possible to perform a large range of computations, with special focus on AI-based tasks on privacy-sensitive data, locally at the edge. This makes it possible to only rely on a backend cloud for communicating results of non-private data, for example by aggregation, or performing computations requiring the combination and further processing of results from different edge devices or historical data sources as it is done in federated learning approaches. In addition, this increase in computation capabilities also enables neighboring edge devices to perform tasks collaboratively, leveraging on each other's available computational resources, before offloading data and computations to a cloud backend.

This workshop will focus on AI and ML techniques related to edge computing systems, and security and privacy approaches in view of data sharing in order to enable the smart and sustainable planning and operation of resource constrained IoT and edge computing applications. The workshop is organized within the scope of the European ITEA3 MIRAI R&D project. It welcomes innovative contributions, early results and position papers addressing one or more of to the topics listed below and intends to foster informal discussions and cross-fertilization on the convergence of AI and edge computing.

Topics

  • Scalability of edge/fog computing IoT applications
  • Run-time adaptation of edge/fog computing infrastructures
  • Advanced AI algorithms and techniques for continuous learning at the edge
  • Distributed and composable ML models and techniques guaranteeing high-quality decision-making
  • Security and privacy on the edge including secure data sharing between edge nodes
  • Low-latency IoT applications deployment and operation on edge computing platforms
  • Benchmarking of AI solutions for edge computing
  • Industrial experiences/show cases for AI enabled edge computing platform management, orchestration, operation
  • Industrial experiences and show cases of distributed AI applications running on edge computing infrastructures
  • Real-world implementations of on-demand computation infrastructures
  • Development and refinement of AI/ML algorithms to execute in resource constrained devices with example cases in predictive maintenance
  • Distributed management and orchestration of IoT applications at edge infrastructure including federated learning approaches

Submission rules

  • Authors should submit their papers as Postscript, PDF or MSWord files.
  • The total length of a paper should not exceed 10 pages IEEE style (including tables, figures and references). IEEE style templates are available here.
  • Papers will be refereed and accepted on the basis of their scientific merit and relevance to the workshop.
  • Preprints containing accepted papers will be published on a USB memory stick provided to the FedCSIS participants.
  • Only papers presented at the conference will be published in Conference Proceedings and submitted for inclusion in the IEEE Xplore® database.
  • Conference proceedings will be published in a volume with ISBN, ISSN and DOI numbers and posted at the conference WWW site.
  • Conference proceedings will be submitted for indexation according to information here.
  • Organizers reserve right to move accepted papers between FedCSIS technical sessions.
  • The workshop organizers are in discussion with different journals for a special issue.
hrule

FedCSIS 2023 is organized by

Strategic partner