Looking for help to understand how to implement a real life problem with kubeFATE


joana97santos@...
 

Hey, I am Joana, I am doing my master thesis at the RWTH (Aachen, Germany) and I am thinking about using Federated Learning applied to maintenance prediction for shop floor machines and found kubeFATE to be the best option for implementation at the moment. I'm not a kubernetes expert, but I've been studying about it and have already done the example implementation with two parties on minikube but I still have doubts about where the connection to my database is made and at what moment I can choose the ML model used. If someone is interested in helping me understand how to apply FATE concepts in kubeFATE I would be very grateful and would be happy to write a Wiki to help future users who are interested in this subject.

Looking forward to the answers,

Joana.


bingquan.hbq@...
 

Hi Joana,

For the database part of your long question I am not quite sure, but for the ML model to be chosen, there are two configurations related to this when you deploy a model: "model_id" and "model_version".

regards,
Gavin