Experimental hybrid quantum-classical reinforcement learning by boson sampling: how to train a quantum cloner

Area of Science:

• Quantum Information Science
• Machine Learning
• Quantum Computing

Background:

• Quantum gates are fundamental operations in quantum computing.
• Optimal quantum cloning is a key challenge in quantum information processing.
• Hybrid approaches combining machine learning and quantum systems are emerging.

Purpose of the Study:

• To experimentally implement a machine-learned quantum gate.
• To achieve optimal phase-covariant cloning using reinforcement learning.
• To demonstrate the practical feasibility of hybrid quantum-classical machine learning.

Main Methods:

• Experimental implementation of a quantum gate controlled classically.
• Utilizing reinforcement learning with cloning fidelity as the reward.
• Leveraging a setup equivalent to boson sampling for quantum information processing.

Main Results:

• The machine-learned quantum gate achieved nearly optimal cloning fidelity.
• Demonstrated the feasibility of hybrid machine learning for quantum control.
• The system's quantum information processing is analogous to boson sampling.

Conclusions:

• Hybrid machine learning offers a practical approach for controlling quantum systems.
• The experimental setup shows potential for achieving quantum supremacy.
• This work validates the integration of AI with quantum technologies.