Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Supercharging Federated Learning with Flower and NVIDIA FLARE

Holger R. Roth, Daniel J. Beutel, Yan Cheng, Javier Fernandez Marques, Heng Pan, Chester Chen, Zhihong Zhang, Yuhong Wen, Sean Yang, Isaac, Yang, Yuan-Ting Hsieh, Ziyue Xu, Daguang Xu, Nicholas D. Lane, Andrew Feng

TL;DR

The paper addresses interoperability between two leading FL ecosystems, Flower and NVIDIA FLARE, to enable cross-framework production deployment. It proposes an integration that routes Flower's gRPC-based communication through FLARE so Flower projects can run unmodified inside the FLARE runtime. Initial experiments demonstrate reproducible results and show a path to richer monitoring through FLARE's experiment tracking in a hybrid setup. The work broadens the FLARE ecosystem by marrying Flower's ease of use and community with FLARE's production-grade features, setting the stage for more scalable, interoperable federated deployments across institutions.

Abstract

Several open-source systems, such as Flower and NVIDIA FLARE, have been developed in recent years while focusing on different aspects of federated learning (FL). Flower is dedicated to implementing a cohesive approach to FL, analytics, and evaluation. Over time, Flower has cultivated extensive strategies and algorithms tailored for FL application development, fostering a vibrant FL community in research and industry. Conversely, FLARE has prioritized the creation of an enterprise-ready, resilient runtime environment explicitly designed for FL applications in production environments. In this paper, we describe our initial integration of both frameworks and show how they can work together to supercharge the FL ecosystem as a whole. Through the seamless integration of Flower and FLARE, applications crafted within the Flower framework can effortlessly operate within the FLARE runtime environment without necessitating any modifications. This initial integration streamlines the process, eliminating complexities and ensuring smooth interoperability between the two platforms, thus enhancing the overall efficiency and accessibility of FL applications.

Supercharging Federated Learning with Flower and NVIDIA FLARE

TL;DR

The paper addresses interoperability between two leading FL ecosystems, Flower and NVIDIA FLARE, to enable cross-framework production deployment. It proposes an integration that routes Flower's gRPC-based communication through FLARE so Flower projects can run unmodified inside the FLARE runtime. Initial experiments demonstrate reproducible results and show a path to richer monitoring through FLARE's experiment tracking in a hybrid setup. The work broadens the FLARE ecosystem by marrying Flower's ease of use and community with FLARE's production-grade features, setting the stage for more scalable, interoperable federated deployments across institutions.

Abstract

Several open-source systems, such as Flower and NVIDIA FLARE, have been developed in recent years while focusing on different aspects of federated learning (FL). Flower is dedicated to implementing a cohesive approach to FL, analytics, and evaluation. Over time, Flower has cultivated extensive strategies and algorithms tailored for FL application development, fostering a vibrant FL community in research and industry. Conversely, FLARE has prioritized the creation of an enterprise-ready, resilient runtime environment explicitly designed for FL applications in production environments. In this paper, we describe our initial integration of both frameworks and show how they can work together to supercharge the FL ecosystem as a whole. Through the seamless integration of Flower and FLARE, applications crafted within the Flower framework can effortlessly operate within the FLARE runtime environment without necessitating any modifications. This initial integration streamlines the process, eliminating complexities and ensuring smooth interoperability between the two platforms, thus enhancing the overall efficiency and accessibility of FL applications.
Paper Structure (14 sections, 6 figures)

This paper contains 14 sections, 6 figures.

Table of Contents

  1. Keywords
  2. Introduction
  3. Integration Goals
  4. Design Principles
  5. FLARE Multi-Job Architecture
  6. Flower Next
  7. Initial Integration
  8. FLARE Reliable Messaging
  9. Integration Design
  10. Experiments
  11. Integration Without Code Changes
  12. Reproducibility:
  13. Hybrid Integration Using FLARE's Experiment Tracking
  14. Discussion

Figures (6)

  • Figure 1: Flower & NVIDIA FLARE integration.
  • Figure 2: NVIDIA FLARE Multi-job system architecture.
  • Figure 3: Flower SuperLink and SuperNodes.
  • Figure 4: Integration of Flower Apps within FLARE.
  • Figure 5: Comparison of running a Flower application natively (a) or within FLARE (b).
  • ...and 1 more figures