Is it possible to have Federated Learning on Cloud-Edge?

[Martiniann]

Hi everyone,

Currently I am working on a school project about federated learning and came across your framework during exploratory analysis. My project should utilize federated learning in this manner - I have an aggregation server (let's say in a cloud). I want this server to provide model to my 2 Raspberry PIs. These two RPIs would then train the model on a local data for x epochs and provide the trained models/gradients back to the global server. On this server, the results would be federated averaged and new model would be sent to the PIs. Is such a workflow possible with your framework? If so, could you provide me a hint?

Thank you, Best regards

[darien-schettler]

I am also curious about this. Also, if this is not currently possible, what is the estimated timeline for something like this becoming possible within TFF.

[ESWZY]

Hey! Same project as me! I'm also working on it.

Would you have any design ideas? Thx!🥰

[Martin-Stevlik]

Hello, If you are interested in federated learning I can recommend these 2 frameworks : https://github.com/IBM/federated-learning-lib https://github.com/adap/flower

Good luck with your project :)

[ESWZY]

Hello, If you are interested in federated learning I can recommend these 2 frameworks : https://github.com/IBM/federated-learning-lib https://github.com/adap/flower

Good luck with your project :)

Thanks for your kindness help! I will have a try. ❤️

Btw, a good news is I find out a way to implement it based on TFF. 🤣

[ladi-pomsar]

@ESWZY Were you able to use tensorflow federated with clients communicating over the internet?

[ESWZY]

@ESWZY Were you able to use tensorflow federated with clients communicating over the internet?

Actually, yes. We just use TFF as a demo to verify the accessibility between clients and master, in our private network.

For more information, you can use this client and runtime demo to have a try.

[thborba]

@ESWZY How did you do it? I'm trying to make the server communicate with the client and use its data for training, but no progress so far.

[ESWZY]

@ESWZY How did you do it? I'm trying to make the server communicate with the client and use its data for training, but no progress so far.

It needs time to learn, but you can find some examples here: https://github.com/tensorflow/federated/tree/main/tensorflow_federated/examples

[ESWZY]

@thborba Take a look at this example, https://github.com/tensorflow/federated/tree/main/tensorflow_federated/tools

If you get hello world, it means success