Towards a dissipative quantum classifier

TL;DR

This paper introduces a quantum classifier based on dissipative engineering and a central spin-qubit model, enabling classical data encoding in a decoherence-free subspace and demonstrating potential for quantum machine learning.

Contribution

The paper presents a novel dissipative quantum classifier using a central spin-qubit system with tailored dissipation, distinct from standard quantum circuit models.

Findings

Able to prepare arbitrary single-qubit states through training

Derived a classification rule for the dissipative quantum system

Showed potential for efficient quantum machine learning

Abstract

In this paper, we propose a novel quantum classifier utilizing dissipative engineering. Unlike standard quantum circuit models, the classifier consists of a central spin-qubit model. By subjecting the auxiliary qubits to carefully tailored strong dissipations, we establish a one-to-one mapping between classical data and dissipative modes. This mapping enables the encoding of classical data within a decoherence-free subspace, where the central qubit undergoes evolution. The dynamics of the central qubit are governed by an effective Lindblad master equation, resulting in relaxation towards a steady state. We first demonstrate the capability of our model to prepare arbitrary single-qubit states by training the inter-coupling of the system and the external dissipations. By elucidating the underlying classification rule, we subsequently derive a quantum classifier. Leveraging a training set with labeled data, we train the dissipative central spin-qubit system to perform specific classification tasks akin to classical neural networks. Our study illuminates the untapped potential of the dissipative system for efficient and effective classification tasks in the realm of quantum machine learning.