Data

Tools

Indexes

Applications

Experiments

Open Source Science

federations_costs_slicing

This repository is the code basis of the paper intitled "The learning costs of Federated Learning in constrained scenarios"

https://github.com/rgtzths/federations_costs_slicing

Science Score: 57.0%

This score indicates how likely this project is to be science-related based on various indicators:

  • ✓
    CITATION.cff file
    Found CITATION.cff file
  • ✓
    codemeta.json file
    Found codemeta.json file
  • ✓
    .zenodo.json file
    Found .zenodo.json file
  • ✓
    DOI references
    Found 1 DOI reference(s) in README
  • ○
    Academic publication links
  • ○
    Academic email domains
  • ○
    Institutional organization owner
  • ○
    JOSS paper metadata
  • ○
    Scientific vocabulary similarity
    Low similarity (9.4%) to scientific vocabulary

Keywords

deep-neural-networks distributed-learning federated-learning machine-learning
Last synced: 6 months ago · JSON representation ·

Repository

This repository is the code basis of the paper intitled "The learning costs of Federated Learning in constrained scenarios"

Basic Info
  • Host: GitHub
  • Owner: rgtzths
  • License: mit
  • Language: Python
  • Default Branch: main
  • Homepage:
  • Size: 15.4 MB
Statistics
  • Stars: 3
  • Watchers: 1
  • Forks: 0
  • Open Issues: 0
  • Releases: 0
Topics
deep-neural-networks distributed-learning federated-learning machine-learning
Created almost 3 years ago · Last pushed about 2 years ago
Metadata Files
Readme License Citation

README.md

dl_costs

Installing

To run the scenarios locally please install the requirements

pip install -r requirements.txt

If you want to run the experiments in RPIs please follow the rpi_setup_comands.md

If you want to run the experiments in docker please run docker compose up.

Running

To run locally you only need the following commands

mpirun -np 4 python mpi_training.py -d dataset/one_hot_encoding/

or

mpirun -np 4 python mpi_custom_training.py -d dataset/one_hot_encoding/

If you want to run the single_host setting you can run

python model.py

To run on docker you will need to connect to the master container

docker exec -it --user mpiuser dl_costs-master-1 bash

Connect one time to every worker to confirm the fingerprint of the server.

ssh worker1 and then exit

After this you can run the command

mpirun -np 4 -hostfile hostfile python mpi_training.py -d dataset

or

mpirun -np 4 -hostfile hostfile python mpi_custom_training.py -d dataset

inside the code folder.

To change the hyperparameters please confirm the available options in the mpitraining.py and mpicustom_training.py files

Results

The results obtained with the different hyperparameters are presented in the paper_results folder the translation for the folder names is 10g50l -> (decentralized optimization with 10 global epochs and 50 local epochs) and fedsgd_64 -> (centralized optimization with batch size 64)

Authors

License

This project is licensed under the MIT License - see the LICENSE file for details

Citation

If you use this code please site our work: Teixeira, Rafael & Antunes, Mário & Gomes, Diogo & Aguiar, Rui. (2023). The learning costs of Federated Learning in constrained scenarios. 10.1109/FiCloud58648.2023.00011.

Owner

  • Name: Rafael Teixeira
  • Login: rgtzths
  • Kind: user
  • Location: Aveiro
  • Company: Instituto de Telecomunicações

Citation (CITATION.cff) 

# This CITATION.cff file was generated with cffinit.
# Visit https://bit.ly/cffinit to generate yours today!

cff-version: 1.2.0
message: "If you use this software, please cite it as bellow"
authors:
- family-names: "Rafael"
  given-names: "Teixeira"
  orcid: "https://orcid.org/0000-0000-0000-0000"

title: "dl_costs"
version: 1.0.0
doi: 10.1109/FiCloud58648.2023.00011
date-released: 2023-08-16
url: "https://github.com/rgtzths/dl_costs"
preferred-citation:
  type: conference-paper
  authors:
  - family-names: "Teixeira"
    given-names: "Rafael"
    orcid: "https://orcid.org/0000-0001-7211-382X"
  - family-names: "Antunes"
    given-names: "Mário"
    orcid: "https://orcid.org/0000-0002-6504-9441"
  - family-names: "Gomes"
    given-names: "Diogo"
    orcid: "https://orcid.org/0000-0002-5848-2802"
  - family-names: "Aguiar"
    given-names: "Rui L."
    orcid: "https://orcid.org/0000-0003-0107-6253"
  title: "The learning costs of Federated Learning in constrained scenarios"
  doi: 10.1109/FiCloud58648.2023.00011
  conference:
    name: "International Conference on Future Internet of Things and Cloud"
    city: "Marrakesh"
    region: "Marrakesh"
    country: "Marocco"
    date-start: 2023-08-14
    date-end: 2023-08-16

GitHub Events

Total
Last Year

Dependencies

Dockerfile docker
  • tensorflow/tensorflow latest build
docker-compose.yml docker
requirements.txt pypi
  • matplotlib ==3.7.0
  • mpi4py ==3.1.4
  • pandas ==1.4.3
  • scikit-learn ==1.1.2
  • tensorflow ==2.10.0