Skip to content

Robust Global FL Pocket Refs#154

Merged
emersodb merged 8 commits intomainfrom
dbe/robust_global_fl_refs
Jun 4, 2025
Merged

Robust Global FL Pocket Refs#154
emersodb merged 8 commits intomainfrom
dbe/robust_global_fl_refs

Conversation

@emersodb
Copy link
Collaborator

[BOOK]

FL

Type of Change

  • New Pocket Reference
  • Edit to Existing Pocket Reference
  • Other (please describe):

Related Issue

Fixes #

Book

  • fundamentals
  • nlp
  • cv
  • rl
  • fl
  • responsible_ai

Description

Adding in 4 new pocket references to flesh out the Robust Global FL chapter. These include:

  1. A general introduction and motivation for Robust Global FL methods
  2. FedOpt
  3. FedProx
  4. MOON

Checklist

  • I have included appropriate contributor tags ({{#author}} or {{#authors}})
  • I have added the reading time preprocessor tag under the title
  • Content is concise and within 7 minutes reading time
  • I have included relevant references and further reading links
  • I have tested locally using mdbook watch books/<book-name> --open
  • Pre-commit hooks pass without errors
  • I have linked to related issues

Additional Context

@emersodb emersodb requested review from Viky397 and nerdai May 16, 2025 22:12
@nerdai
Copy link
Collaborator

nerdai commented May 16, 2025

wooohoo!

Copy link
Collaborator

@nerdai nerdai left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This is really awesome! I left only a few minor comments/edits, but outside from those this is ready to go!

Perhaps in a future PR, we could summarize these methods in a table (pros/cons, when to use, memory and time complexities).

Concept drift is characterized by a change in \\(\\mathbb{P}(Y \vert X)\\) provided a
fixed \\(\\mathbb{P}(Y)\\). Essentially, this drift encapsulates a shift in the
predictive relationship between the features, \\(X\\), and the labels, \\(Y\\).
As an illustrative example, consider training a purchase conversion model
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think we could use Trump tarrifs as part of this example, but maybe we shouldn't be political.

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I may or may not have been thinking of this when I wrote the example, but chose to be more vague 😂


A number of experiments show that the proposed FedOpt family of algorithms
can outperform FedAvg, especially in heterogeneous settings. Moreover, these
algorithms, in the experiments of the paper, outperform SCAFFOLD,[^6] a
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

should we have a pocket ref for SCAFFOLD in this section as well, eventually?

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yes, probably. It'll be a bit more involved, just because the method is more complicated, but I have all of the figures necessary to do one. If you want to throw it on the backlog, I can tackle it at the same time as I work on pFL methods.

Copy link

@Viky397 Viky397 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Looks great @emersodb ! Thank you for all the details and the figures.

@emersodb
Copy link
Collaborator Author

emersodb commented Jun 4, 2025

@nerdai: Just want to highlight that I also updated the algorithm rendering for FedSGD, which will require a new cloud upload. The reason was that the algorithm was number 2 even though there was no preceding algorithm in the chapter, which was bugging me 😂

@emersodb emersodb merged commit 843f07d into main Jun 4, 2025
1 check passed
@emersodb emersodb deleted the dbe/robust_global_fl_refs branch June 4, 2025 15:09
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants