Incentive Allocation in Vertical Federated Learning Based on Bankruptcy Problem