核心概念
Federated bandit learning with heterogeneous clients improves regret reduction through collaborative clustering.
統計
Die Nachfrage nach kollaborativem und privatem Bandit-Lernen steigt aufgrund der wachsenden Menge an Daten aus verteilten Systemen.
引用
"Almost all previous works rely on strong assumptions of client homogeneity, greatly restricting the application of federated bandit learning in practice."
"Our proposed algorithm achieves non-trivial sub-linear regret and communication cost for all clients."