When to Speak, When to Stay Qubit: How Sporadic Updates Tame Quantum Noise
TL;DR for operators SpoQFL is a proposal for making quantum federated learning less fragile by teaching noisy clients when to speak and when to stay quiet.1 In ordinary federated learning, each client trains locally and sends model updates to a server. In quantum federated learning, those clients are quantum models running under noisy intermediate-scale quantum conditions, which means their updates can be corrupted by gate errors, measurement uncertainty, decoherence, and client-to-client hardware variation. ...