Quantum Algorithms for State Preparation and Data Classification based on Stabilizer Codes

TL;DR

This paper introduces a quantum circuit model inspired by stabilizer codes for data classification, utilizing syndrome detection as a quantum perceptron, and presents a recursive algorithm for efficient quantum state preparation.

Contribution

It proposes a novel quantum perceptron model based on stabilizer codes and introduces RASA, an efficient recursive algorithm for approximate quantum state preparation.

Findings

Numerical demonstration of the quantum perceptron model.

Development of RASA for polynomial-depth quantum state encoding.

Integration of stabilizer codes into quantum neural network architecture.

Abstract

Quantum error correction (QEC) is a way to protect quantum information against noise. It consists of encoding input information into entangled quantum states known as the code space. Furthermore, to classify if the encoded information is corrupted or intact, a step known as syndrome detection is performed. For stabilizer codes, this step consists of measuring a set of stabilizer operators. In this paper, inspired by the QEC approach, and specifically stabilizer codes, we propose a prototype quantum circuit model for classification of classical data. The core quantum circuit can be considered as a \emph{quantum perceptron} where the classification is based on syndrome detection. In this proposal, a quantum perceptron is realized by one stabilizer as part of a stabilizer code, while a quantum neural network (QNN) layer is realized by a stabilizer code which consists of many stabilizers. The concatenation of stabilizer codes results in complex QNNs. The QNN is trained by performing measurements and optimization of a set of parameterized stabilizers. We demonstrate the concept numerically. In this paper we also consider the first challenge to most applications of quantum computers, including data classification, which is to load data into the memory of the quantum computer. This loading amounts to representing the data as a quantum state, i.e., quantum state preparation. An exact amplitude encoding algorithm requires a circuit of exponential depth. We introduce an alternative recursive algorithm which approximates amplitude encoding with only a polynomial number of elementary gates. We name it recursive approximate-scheme algorithm (RASA).