Conformal Risk Sharing: Certified Cost Allocation with Participation Guarantees
Researchers have introduced Conformal Risk Sharing, a novel framework designed to fairly allocate the financial burden of rare adverse events among participants. This method ensures that no individual is made worse off by the arrangement, providing trustworthy caps on future obligations. The approach utilizes split conformal calibration on training and calibration data to generate distribution-free guarantees, validated through experiments on synthetic and real-world datasets. AI
IMPACT Introduces a new method for risk management and cost allocation in data-driven scenarios.