Hybrid quantum learning with data re-uploading on a small-scale superconducting quantum simulator

Abstract

Supervised quantum learning is an emergent multidisciplinary domain bridging between variational quantum algorithms and classical machine learning. Here, we study experimentally a hybrid classifier model accelerated by a quantum simulator - a linear array of four superconducting transmon artificial atoms - trained to solve multilabel classification and image recognition problems. We train a quantum circuit on simple binary and multi-label tasks, achieving classification accuracy around 95%, and a hybrid model with data re-uploading with accuracy around 90% when recognizing handwritten decimal digits. Finally, we analyze the inference time in experimental conditions and compare the performance of the studied quantum model with known classical solutions.

11 pages, 6 figures

References in corpus (6)

• Charge insensitive qubit design derived from the Cooper pair box
• Universal control of a six-qubit quantum processor in silicon
• Beyond Barren Plateaus: Quantum Variational Algorithms Are Swamped With Traps
• Demonstrating a Driven Reset Protocol of a Superconducting Qubit
• Experimental quantum adversarial learning with programmable superconducting qubits
• Avoiding barren plateaus via transferability of smooth solutions in Hamiltonian Variational Ansatz

Cited by in corpus (4)

• Quantum Hamiltonian Embedding of Images for Data Reuploading Classifiers
• Realizing a Continuous Set of Two-Qubit Gates Parameterized by an Idle Time
• Classification and reconstruction for single-pixel imaging with classical and quantum neural networks
• Demonstration of sequential processors with quantum advantage and analysis of classical performance limits