FedCVAE

[jafermarq]

FedCVAE

• Title: Data-Free One-Shot Federated Learning Under Very High Statistical Heterogeneity
• Venue: ICLR 2023
• Link to paper: https://openreview.net/forum?id=_hb4vM3jspB

Do you want to work on this baseline?

🌻 Check everything about the Summer of Reproducibility on flower.dev/summer

All available baselines are listed in the Summer of Reproducibility Dashboard and also in the GitHub Issues with the summer-of-reproducibility label. The content is the same.

📝 It is advised to complete these steps before your start working on your code. But if you can't wait to implement your baseline with Flower (we totally understand it 😄), please ensure you follow the steps on how to contribute a new baseline.

What follows are the steps 1 & 2 in the Summer of Reproducibility instructions.

1. Join the Summer of Reproducibility program

• [x] Join the Flower Slack and say "hi! 👋" in the channel #summer-of-reproducibility.
• [x] Pick a baseline from our curated list <---------------------------------------- [you are doing this now]

  2. Define the scope of your contribution

• [x] What are you going to reproduce? Add a comment to your issue and tell us about your plan regarding this baseline: what experiments from the paper are you reproducing?, for which datasets ? the more details you provide us with the better !

• [x] Check if you are eligible for a reward.

  As we have to comply with US/EU regulations, we have checked that individual contributors based on these countries or territories are eligible: Australia, Austria, Belgium, Bulgaria, Canada, Croatia, Cyprus, Czech Republic, Denmark, Estonia, Finland, France, Germany, Gibraltar, Greece, Hong Kong SAR China, Hungary, India, Ireland, Italy, Japan, Latvia, Liechtenstein, Lithuania, Luxembourg, Malta, Mexico, Netherlands, New Zealand, Norway, Poland, Portugal, Romania, Singapore, Slovakia, Slovenia, South Korea, Spain, Sweden, Switzerland, Thailand, United Arab emirates, United Kingdom United States.

  If where you are based is not on the list, please send us an email (summer@flower.dev) letting us know a bit about yourself (where are you currently based?, are you a university student? do you work at a public institution?). Please tell us the baselines you are interested in implementing (i.e. tell us your GitHub issue if you have crated one). We will reach back to you.

• [x] We will discuss with you about your contribution plan, if it sounds like a substantial enough contribution according to the Summer of Reproducibility rules (check our website flower.dev/summer), you'll get the OK from us to start working on your baseline!

What happens next?

• [x] This item will be moved to the In Progress stage by a member of the Flower Team.

• [ ] Follow the instructions for creating a new baseline which will guide you through the process step-by-step.

Is something wrong or not clear ?

• Ask a question directly in your issue.
• Reach out to us via the Flower Slack and ask your question in the #summer-of-reproducibility channel
• Check all the details (including FAQ) in the Summer of Reproducibility website: flower.dev/summer

[conorhassan]

I want to try and implement FedCVAE-KD and FedCVAE-Ens and replicate the results in the paper that was provided for the MNIST and Fashion-MNIST at the varying statistical heterogeneity ($\alpha$) levels, and the varying $m$ number of clients.

[jafermarq]

Hi @conorhassan, it's great you are interested in implementing FedCVAE using Flower. So your plan would be to reproduce experiments in Tables 2 & 3 for MNIST and Fashion-MNIST? What will you use as baseline? FedAvg?

[conorhassan]

It would be the first two third of tables 1 and 2 (i.e., not the rows corresponding to the SVHN data set). Yes, initially consider FedAvg, but ideally, a nice comparison would be to a method that is a little more tailored to a one-shot approach than FedAvg.

I know I am biased here, but the methods in this paper https://arxiv.org/abs/2302.03314 can be used for one-shot FL and maybe a more competitive comparison than FedAvg. I can hopefully contribute to this in the future.

[jafermarq]

Hi @conorhassan , sorry for the delay getting back to you. Your contribution plan makes sense, let's start with just FedAvg ? Just a quick question I'm asking to all recent contributors: are you confident you have access to the necessary compute resources (i.e. GPUs) to replicate these experiments?

It would be amazing to include your own work as a baseline here too (up to you really). Or why not having it as a standalone baseline after the Summer of Reproducibility? (your results in Table 1 look pretty good)

[conorhassan]

I'm confident that I have access to the necessary compute

[jafermarq]

Ok. Thank you for confirming. Then, FedCVAE is yours to reproduce with Flower. I have now ✅ all points in Steps 1 & 2 above, added you as the assignee to this issue and moved it to In Progress status. You'll find how to start with your baseline by following the guide linked in the What happens next? section above. If you have any doubts (no matter how simple or complex) please reach out to me or other contributors via the Flower Slack.

Looking forward to seeing FedCVAE in action!