Until now, XGB seems the most credible tabular model algorithm.
Also, histogram-based XGboost is a CPU-trainable model algorithm.
And also, when we validate the model in other CDM databases, XGboost will be the easiest way.
I'm not sure that the other PyTorch model is CDM applicable or not
So, I think we should apply Federated XGB to our package
Until now, XGB seems the most credible tabular model algorithm. Also, histogram-based XGboost is a CPU-trainable model algorithm.
And also, when we validate the model in other CDM databases, XGboost will be the easiest way. I'm not sure that the other PyTorch model is CDM applicable or not
So, I think we should apply Federated XGB to our package