Robust and Fair Online Allocation with Unreliable Advice

A new rule combines predictive advice with a conservative fallback and fairness correction to handle unreliable predictions in online allocation. The method maintains stability under adversarial advice and reduces exposure disparity in scenarios with irreversible decisions.