Skip to content

Implemented FedAvg & FedProx: Decentralized Optimization Algorithms for Neural Networks for an Image Classification Task- Distributed Optimization and Learning(DOL) Course Project

Notifications You must be signed in to change notification settings

erfunmirzaei/FedAvg-FedProx

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

6 Commits
 
 
 
 
 
 

Repository files navigation

DOL-Project1

Implemented FedAvg & FedProx: Decentralized Optimization Algorithms for Neural Networks for an Image Classification Task

Overview:

In this project, we implemented federated learning algorithms as samples of the decentralized optimization algorithms for the case when client datasets are i.i.d. or non-iid. Then, compare the results with the centralized version.

For this purpose, a convolution neural network and the cifar10 dataset are used. In the first part, we trained the network in a centralized manner, as in the standard cases in deep learning. We implemented the FedAvg algorithm, one of the most famous algorithms in federated learning, when the client datasets are iid. In the following, the iid assumption was discarded. Thus, we split datasets in a non-iid manner. Therefore for non-iid datasets, we implemented the FedProx algorithm and compared the results to the FedAvg Algorithm.

About

Implemented FedAvg & FedProx: Decentralized Optimization Algorithms for Neural Networks for an Image Classification Task- Distributed Optimization and Learning(DOL) Course Project

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published