Bandit Algorithm Driven by a Classical Random Walk and a Quantum Walk

Quantum walks (QWs) have the property that classical random walks (RWs) do not possess -- coexistence of linear spreading and localization -- and this property is utilized to implement various kinds of applications. This paper proposes a quantum-walk-based algorithm for multi-armed-bandit (MAB) problems by associating the two operations that make MAB problems difficult -- exploration and exploitation -- with these two behaviors of QWs. We show that this new policy based on the QWs realizes high performance compared with the corresponding RW-based one.