Experimental quantum-enhanced kernel-based machine learning on a photonic processor.

Area of Science:

• Quantum computing
• Machine learning
• Photonic integrated circuits

Background:

• Machine learning (ML) demands significant energy and computational resources for complex tasks.
• Quantum computation offers potential for reduced resource requirements, but feasibility with current technology is uncertain.

Purpose of the Study:

• To demonstrate a quantum kernel method for binary classification using a photonic integrated processor.
• To evaluate the performance of the quantum protocol against state-of-the-art classical kernel methods.

Main Methods:

• Implementation of a kernel method on a photonic integrated processor.
• Utilization of quantum interference and single-photon coherence for enhanced computation.
• System dimension modification via additional modes and injected photons, without requiring entangling gates.

Main Results:

• The quantum kernel method outperforms classical kernel methods like Gaussian and Neural Tangent Kernels.
• Single-photon coherence further improves classification accuracy.
• The scheme demonstrates a viable approach for quantum-enhanced machine learning on current hardware.

Conclusions:

• Quantum effects can significantly improve standard machine learning algorithms.
• This work provides a pathway towards more efficient quantum algorithms for complex computational tasks.
• Photonic quantum processors offer a promising platform for practical quantum machine learning applications.