Quantum Reinforcement Learning

By José D. Martín-Guerrero I Lucas Lamata 

Inside QML, the field of QRL has raised an increasing interest in recent years [11]. Here, the aim is to develop ‘‘intelligent” quantum agents, which may interact with the outer world and learn from it, in order to achieve some specified goal. In this sense, several works have made interesting proposals in the past few years [27,31,51,25,52–55]. Some of these works deal with quantum agents interacting via Grover search with a classical environment [27,31], others deal with quantum agents coupled to an oracular quantum environment, with proved quantum speedup [51], while some other results are related to possible implementations of quantum agents interacting with few-qubit quantum environments [25,52–55]. In this sense, it is remarkable an experiment of quantum reinforcement learning with quantum photonics [54], in which a speedup was demonstrated with respect to standard quantum state tomography, in the limit of a small amount of resources available.

Read more