wrong paper associated with FedXgbNnAvg strategy in the docs about flower

[gfmota]

I can help with it. Can I be assiigned?

[jafermarq]

Hey @gfmota, i've just assigned this issue to you.

I've discussed about this briefly with @panh99 , and we think this can easily be fixed if the baseline @Aml-Hassan-Abd-El-hamid mentions had a __init__ method implemented (with docstrings pointing to the paper where that first proposed the strategy). This issue does not only affect FedXgbNnAvg but also some other strategies that directly inherit from FedAvg w/o defining a __init__. Could you look into all those? You'll likely find the paper to cite at the top of the respective .py file.

One way of doing the above would look like this. But we are open to other ways of doing it:

class FedXgbNnAvg(FedAvg):
    """Configurable FedXgbNnAvg strategy implementation."""
    def __init__(self, *args, **kwargs):
        """Federated XGBoost [Ma et al., 2023 """ # Edit with the full reference etc
        super().__init__(*args, **kwargs)

[gfmota]

I've created a PR adding this doc in the class, could anyone look it to see if is it what was expected?

[jafermarq]

I've created a PR adding this doc in the class, could anyone look it to see if is it what was expected?

Yes, that seems to do the fix. There are a few other strategies that suffer from the same issue actually. (FedMedian, QFedAvg, FaultTolerantFedAvg at least these). But i think these could be left for another PR?