# Decentralized Federated Learning

**URL:** <https://discuss.flower.ai/t/decentralized-federated-learning/966>\
**Category:** General\
**Tags:** flower\
**Created:** [May 18, 2025, 9:20am UTC](https://discuss.flower.ai/t/decentralized-federated-learning/966 "2025-05-18T09:20:51Z")\
**Posts on this page:** 4\
**Page:** 1

<div class="post-metadata">

**Author:** ![zinebabbazi](https://avatars.discourse-cdn.com/v4/letter/z/bbce88/32.png) [@zinebabbazi](https://discuss.flower.ai/u/zinebabbazi)\
**Post date:** [May 18, 2025, 9:20am UTC](https://discuss.flower.ai/t/decentralized-federated-learning/966/1 "2025-05-18T09:20:51Z")

</div>

Hello everyone,

I’m curious if there have been any recent advances or new trends in Peer-to-Peer (P2P) communication for Decentralized Federated Learning, especially using the Flower framework or similar tools.

This is a key topic for my research, and I’m looking for protocols or libraries that support effective P2P communication in distributed learning setups — whether experimental or practical.

Any pointers or recommendations would be greatly appreciated!

---

<div class="post-metadata">

**Author:** ![tijnvdheuvel](https://avatars.discourse-cdn.com/v4/letter/t/e0b2c6/32.png) [@tijnvdheuvel](https://discuss.flower.ai/u/tijnvdheuvel)\
**Post date:** [May 23, 2025, 7:09am UTC](https://discuss.flower.ai/t/decentralized-federated-learning/966/2 "2025-05-23T07:09:25Z")

</div>

Hi Zinebabbazi,

I am working on a research paper also involving P2P decentralized learning. If this works out using flower I will also update you on what worked for me.  
I am trying to have the paper done around the end of June 2025.

Cheers!  
Tijn

---

<div class="post-metadata">

**Author:** ![blueberry](https://avatars.discourse-cdn.com/v4/letter/b/54ee81/32.png) [@blueberry](https://discuss.flower.ai/u/blueberry)\
**Post date:** [May 24, 2025, 5:48am UTC](https://discuss.flower.ai/t/decentralized-federated-learning/966/3 "2025-05-24T05:48:12Z")

</div>

Hi Zinebabbazi,

I am working on an experimental project called the “MyceliumWebServer”, which basically builds on federated learning but also allows nodes to switch learning groups: [GitHub - bluebbberry/MyceliumWebServer: Federated learning based on ActivityPub (a federated social networking protocol).](https://github.com/bluebbberry/MyceliumWebServer). I already got a working prototype implemented, but it has huge scaling issues and it does not seem to be combined with flower ai.

Hope this helps

Best,  
Jan

---

<div class="post-metadata">

**Author:** ![williamlm](https://dub1.discourse-cdn.com/flex013/user_avatar/discuss.flower.ai/williamlm/32/162_2.png) [@williamlm](https://discuss.flower.ai/u/williamlm)\
**Post date:** [August 5, 2025, 6:31pm UTC](https://discuss.flower.ai/t/decentralized-federated-learning/966/4 "2025-08-05T18:31:10Z")

</div>

hi @zinebabbazi,

there’s currently an open PR for this: [feat(framework) Add serverless federated learning support by zzsi · Pull Request #5189 · adap/flower · GitHub](https://github.com/adap/flower/pull/5189), feel free to add your thoughts and ideas on how we can merge this PR.

It is quite lengthy, so maybe better to break it down into sub-components?