Comparing quantum hybrid reinforcement learning to classical methods

Abstract In recent history, reinforcement learning (RL) proved its capability by solving complex decision problems mastering several games. Increased computational power and the advances in approximation with neural networks (NN) paved path to RL’s successful applications. Even though RL can tackle more nowadays, it still relies on runtime. Quantum computing promises solve these issues encode information potential quadratic speedup We compare tabular Q-learning using either a quantum or classical architecture frozen lake problem. Furthermore, three algorithms are analyzed terms of iterations until convergence optimal behavior, memory usage, Within paper, NNs utilized for domain, while domain variational circuits, as hybrid method, have been used. Our simulations show that approximator is beneficial usage provides better sample complexity than NNs; however, lacks speed be competitive.