1.Implement Generated Algorithm to choose next generation 2.Using Federated Learning to calculated two objects(f1: test error, f2: communication round), which are input in generated learning
Step 1: GenAlgrithm choose R population by random
Step 2: Do t times communication in Federated Learning
Step 3: Calculate R numbers of f(f1, f2)
Step 4: Choose M numbers of Perato Optimal solution, can keep record their hyperparameters
Step 5: Reapet the previous process till the result we find final M results
Step 6: Choose the best performanced point and using their hyperparameter as our model's hyperparameter.