Computing with two quantum reservoirs connected via optimized two-qubit nonselective measurements

TL;DR

This paper introduces a novel quantum reservoir network using optimized two-qubit measurements and purification channels, significantly improving information transfer and computational performance in quantum machine learning systems.

Contribution

It proposes a new method to connect quantum reservoirs with optimized measurements and channels, enhancing scalability and performance in quantum reservoir computing.

Findings

Efficient information transfer in a 7-qubit network with performance comparable to larger classical-linked networks.

Optimized channels improve quantum reservoir computing performance.

The approach overcomes size and noise limitations in quantum systems.

Abstract

Currently, quantum reservoir computing is one of the most promising and experimentally accessible techniques for hybrid, quantum-classical machine learning. However, its applications are limited due to practical restrictions on the size of quantum systems and the influence of noise. Here we propose a novel approach to connect two quantum reservoirs in a network to overcome these issues and enhance their computing performance. To transfer information between quantum reservoirs, we perform optimized two-qubit non-selective measurements. We suggest a general heuristic optimization strategy based on tensor network language and matrix representation of two-qubit quantum channels specified for quantum reservoir computing. In addition, we introduce a single qubit purification channel and its optimization for further enhancement of quantum reservoir computing. We also demonstrate that the optimized channels applied to a seven-qubit network can efficiently transfer information between its parts with the resulting performance comparable to a network up to twenty-five qubits connected via a classical information link.