Machine Learning for Edge-Cloud Systems (ML4ECS)
CfP

In conjunction with HiPEAC 2025

HiPEAC 2025 workshop
January 22, 2025
Barcelona, Spain

Autonomic end-to-end system management across the cloud-edge continuum is becoming increasingly critical as computational demands, scale, dynamics, and system heterogeneity grow rapidly. Machine Learning for Edge-Cloud Systems (ML4ECS) is a full-day workshop that focuses on academic and industrial innovation and research on developing and deploying AI/ML models to enable optimized system management and application execution on the continuum.

The workshop aims to bring together a community of researchers and practitioners who study problems at the intersection of ML, Edge-Cloud continuum, distributed and dependable system operation, and flexible, adaptive, and resilient application deployment. It seeks to reach a broader community invested in this fast-evolving field, aiming to increase awareness of ongoing initiatives, encourage collaboration, and promote the open exchange of ideas.

Call For Papers

The workshop invites the research community to submit expressions of interest to provide a talk based on an extended abstract format focusing on topics such as (but not limited to):

  • ML models to improve overall system performance metrics (such as latency, throughput, energy dissipation, use of green energy, security, etc.) across the Edge-Cloud continuum.
  • ML models for efficient and transparent pooling of edge resources, seamless integration with cloud and edge infrastructures, resource configuration optimization, multi-level orchestration, workload migration, security, and anomaly detection.
  • Privacy-preserving decentralized learning mechanisms and adaptive networking services.
  • Learning mechanisms for sustainable computing with lower carbon intensity in Edge-Cloud operations.
  • ML-based Open-source tools and frameworks for large-scale experimentation in edge systems.
  • Multi-cluster, multi-domain environments and applications.
  • Network communication and optimization across resource-restricted edge environments.
  • AI/ML-enabled tools and mechanisms in resource-constrained environments.
  • AI/ML-based Joint orchestration of cross-layer data (network, compute, data resources).

Papers will be evaluated by the workshop’s technical program committee based on the quality of the submission, its relevance to the workshop’s topics, and its potential to ignite discussions about future directions, and solutions related to the mentioned topics. We welcome research papers, case studies, and position papers, as well as demos.

Moreover, we will organize panel discussions at the end of each technical session to allow for more comprehensive exploration of complex topics and drive active participation from the audience through Q&As.

The workshop will present a Best Talk award, sponsored by the HE CODECO project.

There will be no formal proceedings of the workshop. The authors are free to include the content of the extended abstract in a paper submitted to a journal or conference with proceedings. The presentations will be made publicly available via this website and/or HiPEAC. 

Important Dates

Extended Abstract submission: October 30 November 6, 2024, 23:59 AoE

Notification: November 10 November 15, 2024

Workshop: January 22, 2025

Paper Format

Papers should be in double column IEEE format  up to 2 pages, not including references.

Papers should be uploaded as PDF.

Submissions are double blind.

Submission Site

Submissions will be made at easychair.org.

Program

8:15 – 9:45
HiPEAC Conference Welcome and Keynote

10:00 – 11:00
ML4ECS Keynote Speech

Danh Le Phuoc (Technical University Berlin)

Title: Edge-driven MLOps via Computing Swarm

Abstract: This talk will share my view on Edge-Driven MLOps through the lens of Computing Swarm. The first part introduces the concept of Computing Swarm, emphasizing its role in building highly dynamic and distributed ML-based systems such as collaborative robots, software defined vehicles and V2X/I2V across the edge-cloud continuum. Following this, I will delve into the modern toolboxes and the design space tailored for Computing Swarm, with real-world use cases inspired by Horizon Europe and Chip-JU projects such as SmartEdge, SMARTY, and HAL4SDV. I then showcase our vision and current work for harnessing the full potential of computing resources at the network edge, enabling scalable, efficient, and intelligent applications. The talk concludes by discussing key challenges and outlining future directions for advancing the swarm computing paradigm amid the recent fast development of AI.

11:00 – 11:30
Coffee Break

11:30 – 13:00
ML-based Edge-Cloud management

  • 11:30 – 11:45: Alexandros Patras, Nikolaos Bellas, Christos D. Antonopoulos and Spyros Lalis. Managing Computational Resources with Machine Learning Policies
  • 11:45 – 12:00: Ferran Diego Andilla and Francisco Alvarez Terribas. Scheduling Inference Workloads on Distributed Edge Clusters with Reinforcement Learning
  • 12:00 – 12:15: Isaac David NĂșñez Araya, Michael Gerndt and Mohak Chadha. EdgeLessPart: Distributed and Progressive Inference for the Edge-Cloud Continuum
  • 12:15 – 12:30: Georgia Christofidi, Francisco Alvarez Terribas, Jesus Alberto Omaña Iglesias, Nicolas Kourtellis and Thaleia Dimitra Doudali.CaRE: Towards Carbon and Resource Efficient Orchestration at the Cloud-Edge Continuum
  • 12:30 – 12:35: Claudio Cicconetti. Benchmarking an EDGELESS Cluster for Serverless Edge Computing Applications.
  • 12:35 – 13:00: Panel of paper authors

13:00 – 14:00
Lunch

14:00 – 15:30
ML4ECS Demo Session

  • 14:00 – 14:15: Georgio Koukis, Lefteris Mamatas, Vasilis Tsaoussidis (ATHINA) and Panagiotis Karamolegkos (University of Piraeus Research Center). The CODECO Experimentation Framework
  • 14:15 – 14:30: Alejandro Espinosa, Rizkallah Touma. Scheduling applications across the Edge-Cloud based on AI recommendations: the CODECO PDLC approach
  • 14:30 – 14:45: Dalal Ali, Rute C. Sofia. Bringing energy-awareness to the Edge-Cloud orchestration
  • 14:45 – 15:12: F. Pournaropoulos et al. Adaptive Application and System Management Across the Mobile-Edge-Cloud Continuum: A Smart Agriculture Use Case
  • 15:13 – 15:30: Álvaro Curto Merino and Claudio Cicconetti. Benchmarking and Run-time Data Analysis

15:30 – 16:00
Coffee Break

16:00 – 17:30
Applications and System-level mechanisms

  • 16:00 – 16:15: Mbasa Joaquim Molo, Lucia Vadicamo, Emanuele Carlini, Claudio Gennaro and Richard Connor. Experimenting with Information Dissimilarity for Knowledge Distillation at the Edge
  • 16:15 – 16:30: Andrea Pazienza and Massimiliano Rossi. Reinforcement Learning Training Strategies for 5G Networks Latency Optimization
  • 16:30 – 16:45: John Byabazaire and Dimitris Chatzopoulos. MLaaS: Decoupling Application Intelligence from Application Logic
  • 16:45 – 17:00: Anastassios Nanos, Charalampos Mainas, Georgios Ntoutsos and Ilias Lagomatis. Scalable and Lightweight Cloud-Native Application Sandboxing
  • 17:00 – 17:05: Marcell Feher and John Byabazaire. A Dynamic and Cost-Efficient Multi-Region Object Storage System Using Erasure Coding
  • 17:05 – 17:30: Panel of paper authors

17.30
End of Workshop

Workshop Organizers

Chair: Nikolaos Bellas (University of Thessaly, EL, and MLSysOps)

  • Rute C. Sofia (fortiss GmbH and HE CODECO Coordinator)
  • Raffaele Gravina (University of Calabria, IT, and MLSysOps)
  • Jesus Omaña Iglesias (TelefĂłnica and EDGELESS)

Technical Program Committee:

    • Ana Solange (INOVA +, PT)
    • Josep Martrat (ATOS, ES)
    • Vasileios Theodorou (Intracom-Telecom, EL)
    • Vassilis Tsaoussidis (ATHENA, EL)
    • Xiaoming Fu (University of Göttingen, DE)
    • Harald MĂŒller (Siemens AG, DE)
    • John Soldatos (Netcompany-Intrasoft, LU)
    • Marco Jahn (Eclipse Foundation, DE)
    • Rizkallah Touma (I2CAT, ES)
    • Dimosthenis Kyriazis (University of Piraeus Research Center, EL)
    • Luis Miguel Contreras (Telefonica, ES)
    • David JimĂ©nez (Universidad Politecnica de Madrid, ES)
    • Andries Stam (Almende)
    • Josh Salomon (RedHat)
    • Luis GarcĂ©s-Erice (IBM)
    • Adel N. Toosi (University of Melbourne, AU)
    • Borja Saez (Infineon, DE)
    • Carlos Molina (Eurecat, ES)
    • Chen Chen (University of Cambridge, UK)
    • Emanuele Carlini (ISTI-CNR, IT)
    • Francisco Vicente (Worldline Iberia SA, ES)
    • JesĂșs Omaña Iglesias (Telefonica Research, ES)
    • Manos Athanatos (Technical University of Crete, EL)
    • Markus Sauer (Siemens AG, DE)
    • Raphael Hetzel (Technical University of Munich, DE)
    • Roberto Riggio (UniversitĂ  Politecnica delle Marche, IT)
    • Roman Kolcun (University of Cambridge, UK)
    • ​​Timothy Wood (George Washington University, USA)
    • Nikolaos Bellas (University of Thessaly, EL)
    • Christos D. Antonopoulos (University of Thessaly, EL)
    • Spyros Lalis (University of Thessaly, EL)
    • Raffaele Gravina (University of Calabria, IT)
    • Gianluca Aloi (University of Calabria, IT)
    • Dimitrios Chatzopoulos (University College Dublin, IE)
    • John Byabazaire (University College Dublin, IE)
    • Anastasios Nanos (NUBIS, EL)
    • Christos Panagiotou (NUBIS, EL)
    • Marcel Feher (Chocolate Cloud, DK)
    • Bowen Sun (University of Thessaly, EL)
    • Alexandros Patras (University of Thessaly, EL)

Contact

If you have any questions, please feel free to send an email to Nikolaos Bellas at nbellas@uth.gr.