On the Resource Consumption of Distributed ML
The convergence of Machine Learning (ML) with the edge computing paradigm has paved the way for distributing processing-heavy ML tasks to the network's extremes. As the edge deployment details still remain an open issue, distributed ML schemes tend to be network-agnostic; thus, their effect on...
Published in: | 2021 IEEE International Symposium on Local and Metropolitan Area Networks (LANMAN) |
---|---|
Main Authors: | , , , , |
Format: | Report |
Language: | English |
Published: |
Zenodo
2021
|
Subjects: | |
Online Access: | https://doi.org/10.1109/LANMAN52105.2021.9478809 |
id |
ftzenodo:oai:zenodo.org:6861384 |
---|---|
record_format |
openpolar |
spelling |
ftzenodo:oai:zenodo.org:6861384 2024-09-15T18:03:52+00:00 On the Resource Consumption of Distributed ML Georgios Drainakis Panagiotis Pantazopoulos Konstantinos Katsaros Vasilis Sourlas Angelos Amditis 2021-07-12 https://doi.org/10.1109/LANMAN52105.2021.9478809 eng eng Zenodo https://zenodo.org/communities/5g_iana https://zenodo.org/communities/eu https://doi.org/10.1109/LANMAN52105.2021.9478809 oai:zenodo.org:6861384 info:eu-repo/semantics/openAccess Creative Commons Attribution 4.0 International https://creativecommons.org/licenses/by/4.0/legalcode IEEE LANMAN, IEEE International Symposium on Local and Metropolitan Area Networks, 12-14 July 2021 #machinelearning #ML #DML # edgenetwork #edgecomputing info:eu-repo/semantics/preprint 2021 ftzenodo https://doi.org/10.1109/LANMAN52105.2021.9478809 2024-07-27T02:52:06Z The convergence of Machine Learning (ML) with the edge computing paradigm has paved the way for distributing processing-heavy ML tasks to the network's extremes. As the edge deployment details still remain an open issue, distributed ML schemes tend to be network-agnostic; thus, their effect on the underlying network's resource consumption is largely ignored.In our work, assuming a network tree structure of varying size and edge computing characteristics, we introduce an analytical system model based on credible real-world measurements to capture the end-to-end consumption of ML schemes. In this context, we employ an edge-based (EL) and a federated (FL) ML scheme and in-depth compare their bandwidth needs and energy footprint against a cloud-based (CL) baseline approach. Our numerical evaluation suggests that EL exhibits a minimum of 25% bandwidth-efficiency compared to CL and FL, if employed by a few nodes higher in the edge network, while halving the network's energy costs. Report DML Zenodo 2021 IEEE International Symposium on Local and Metropolitan Area Networks (LANMAN) 1 6 |
institution |
Open Polar |
collection |
Zenodo |
op_collection_id |
ftzenodo |
language |
English |
topic |
#machinelearning #ML #DML # edgenetwork #edgecomputing |
spellingShingle |
#machinelearning #ML #DML # edgenetwork #edgecomputing Georgios Drainakis Panagiotis Pantazopoulos Konstantinos Katsaros Vasilis Sourlas Angelos Amditis On the Resource Consumption of Distributed ML |
topic_facet |
#machinelearning #ML #DML # edgenetwork #edgecomputing |
description |
The convergence of Machine Learning (ML) with the edge computing paradigm has paved the way for distributing processing-heavy ML tasks to the network's extremes. As the edge deployment details still remain an open issue, distributed ML schemes tend to be network-agnostic; thus, their effect on the underlying network's resource consumption is largely ignored.In our work, assuming a network tree structure of varying size and edge computing characteristics, we introduce an analytical system model based on credible real-world measurements to capture the end-to-end consumption of ML schemes. In this context, we employ an edge-based (EL) and a federated (FL) ML scheme and in-depth compare their bandwidth needs and energy footprint against a cloud-based (CL) baseline approach. Our numerical evaluation suggests that EL exhibits a minimum of 25% bandwidth-efficiency compared to CL and FL, if employed by a few nodes higher in the edge network, while halving the network's energy costs. |
format |
Report |
author |
Georgios Drainakis Panagiotis Pantazopoulos Konstantinos Katsaros Vasilis Sourlas Angelos Amditis |
author_facet |
Georgios Drainakis Panagiotis Pantazopoulos Konstantinos Katsaros Vasilis Sourlas Angelos Amditis |
author_sort |
Georgios Drainakis |
title |
On the Resource Consumption of Distributed ML |
title_short |
On the Resource Consumption of Distributed ML |
title_full |
On the Resource Consumption of Distributed ML |
title_fullStr |
On the Resource Consumption of Distributed ML |
title_full_unstemmed |
On the Resource Consumption of Distributed ML |
title_sort |
on the resource consumption of distributed ml |
publisher |
Zenodo |
publishDate |
2021 |
url |
https://doi.org/10.1109/LANMAN52105.2021.9478809 |
genre |
DML |
genre_facet |
DML |
op_source |
IEEE LANMAN, IEEE International Symposium on Local and Metropolitan Area Networks, 12-14 July 2021 |
op_relation |
https://zenodo.org/communities/5g_iana https://zenodo.org/communities/eu https://doi.org/10.1109/LANMAN52105.2021.9478809 oai:zenodo.org:6861384 |
op_rights |
info:eu-repo/semantics/openAccess Creative Commons Attribution 4.0 International https://creativecommons.org/licenses/by/4.0/legalcode |
op_doi |
https://doi.org/10.1109/LANMAN52105.2021.9478809 |
container_title |
2021 IEEE International Symposium on Local and Metropolitan Area Networks (LANMAN) |
container_start_page |
1 |
op_container_end_page |
6 |
_version_ |
1810441316243341312 |