Trade-offbetweenPayoffandModelRewardsin Shapley-FairCollaborativeMachineLearning