Code for paper "Joint Optimization in Edge-Cloud Continuum for Federated Unsupervised Person Re-identification"

Code for paper "Joint Optimization in Edge-Cloud Continuum for Federated Unsupervised Person Re-identification"
Abstract: Person re-identification (ReID) aims to re-identify a person from non-overlapping camera views. Since person ReID data contains sensitive personal information, researchers have adopted federated learning, an emerging distributed training method, to mitigate the privacy leakage risks. However, existing studies rely on data labels that are laborious and time-consuming to obtain. We present FedUReID, a federated unsupervised person ReID system to learn person ReID models without any labels while preserving privacy. FedUReID enables in-situ model training on edges with unlabeled data. A cloud server aggregates models from edges instead of centralizing raw data to preserve data privacy. Moreover, to tackle the problem that edges vary in data volumes and distributions, we personalize training in edges with joint optimization of cloud and edge. Specifically, we propose personalized epoch to reassign computation throughout training, personalized clustering to iteratively predict suitable labels for unlabeled data, and personalized update to adapt the server aggregated model to each edge. Extensive experiments on eight person ReID datasets demonstrate that FedUReID not only achieves higher accuracy but also reduces computation cost by 29%. Our FedUReID system with the joint optimization will shed light on implementing federated learning to more multimedia tasks without data labels.

EasyFL: A Low-code Federated Learning Platform

PyPI docs license maintained Downloads

📘 Documentation | 🛠️ Installation

Introduction

EasyFL is an easy-to-use federated learning (FL) platform based on PyTorch. It aims to enable users with various levels of expertise to experiment and prototype FL applications with little/no coding.

You can use it for:

  • FL Research on algorithm and system
  • Proof-of-concept (POC) of new FL applications
  • Prototype of industrial applications
  • Learning FL implementations

We currently focus on horizontal FL, supporting both cross-silo and cross-device FL. You can learn more about federated learning from these resources.

Major Features

Easy to Start

EasyFL is easy to install and easy to learn. It does not have complex dependency requirements. You can run EasyFL on your personal computer with only three lines of code (Quick Start).

Out-of-the-box Functionalities

EasyFL provides many out-of-the-box functionalities, including datasets, models, and FL algorithms. With simple configurations, you simulate different FL scenarios using the popular datasets. We support both statistical heterogeneity simulation and system heterogeneity simulation.

Flexible, Customizable, and Reproducible

EasyFL is flexible to be customized according to your needs. You can easily migrate existing CV or NLP applications into the federated manner by writing the PyTorch codes that you are most familiar with.

Multiple Training Modes

EasyFL supports standalone training, distributed training, and remote training. By developing the code once, you can easily speed up FL training with distributed training on multiple GPUs. Besides, you can even deploy it to Kubernetes with Docker using remote training.

Getting Started

You can refer to Get Started for installation and Quick Run for the simplest way of using EasyFL.

For more advanced usage, we provide a list of tutorials on:

Projects & Papers

We have released the source code for the following papers under the applications folder:

The following publications are developed using EasyFL:

  • Divergence-aware Federated Self-Supervised Learning, ICLR'2022. [paper]
  • Collaborative Unsupervised Visual Representation Learning From Decentralized Data, ICCV'2021. [paper]
  • Joint Optimization in Edge-Cloud Continuum for Federated Unsupervised Person Re-identification, ACMMM'2021. [paper]

💡 We will release the source codes of these projects in this repository. Please stay tuned.

We have been doing research on federated learning for several years, the following are our additional publications.

  • EasyFL: A Low-code Federated Learning Platform For Dummies, IEEE Internet-of-Things Journal. [paper]
  • Performance Optimization for Federated Person Re-identification via Benchmark Analysis, ACMMM'2020. [paper]
  • Federated Unsupervised Domain Adaptation for Face Recognition, ICME'22. [paper]

Join Our Community

Please join our community on Slack: easyfl.slack.com

We will post updated features and answer questions on Slack.

License

This project is released under the Apache 2.0 license.

Citation

If you use this platform or related projects in your research, please cite this project.

@article{zhuang2022easyfl,
  title={Easyfl: A low-code federated learning platform for dummies},
  author={Zhuang, Weiming and Gan, Xin and Wen, Yonggang and Zhang, Shuai},
  journal={IEEE Internet of Things Journal},
  year={2022},
  publisher={IEEE}
}

Download Source Code

Download ZIP

Paper Preview

Aug 20, 2022