README.md

FedDCD.jl

Implementation of the federated dual coordinate descent (FedDCD) method.

Installation

To install, just call

Pkg.add("https://github.com/ZhenanFanUBC/FedDCD.jl.git")

Get data

We get data from the website of LIBSVM. To download the datasets, just call

mkdir data
cd ./data
wget https://www.csie.ntu.edu.tw/~cjlin/libsvmtools/datasets/binary/rcv1_train.binary.bz2
wget https://www.csie.ntu.edu.tw/~cjlin/libsvmtools/datasets/binary/rcv1_test.binary.bz2
wget https://www.csie.ntu.edu.tw/~cjlin/libsvmtools/datasets/multiclass/mnist.scale.bz2
wget https://www.csie.ntu.edu.tw/~cjlin/libsvmtools/datasets/multiclass/mnist.scale.t.bz2
bzip2 -d rcv1_train.binary.bz2
bzip2 -d rcv1_test.binary.bz2
bzip2 -d mnist.scale.bz2
bzip2 -d mnist.scale.t.bz2

Run FedAvg for toy example.

include("experiments/PrimalMethods.jl")
RunFedAvgAndProx(
    "data/rcv1_train.binary",
    "data/rcv1_train.binary"
    1e-2,
    0.0,
    0.3,
    0.1,
    100,
    "results/toy.txt"
    )

Run FedDCD for toy example.

include("experiments/DualMethods.jl")
RunFedDCD(
    "data/rcv1_train.binary",
    "data/rcv1_train.binary"
    1e-2,
    0.3,
    0.1,
    100,
    "results/toy.txt"
    )

Citing this package

If you use FedDCD.jl for published work, we encourage you to cite the software.

Use the following BibTeX citation:

@article{fan2022dual,
      title={A dual approach for federated learning}, 
      author={Zhenan Fan and Huang Fang and Michael P. Friedlander},
      year={2022},
      eprint={2201.11183},
      archivePrefix={arXiv},
      primaryClass={cs.LG}
}

Credits

FedDCD.jl is developed by Zhenan Fan and Huang Fang