论文标题
实验实施量子算法用于关联规则挖掘
Experimental implementation of quantum algorithm for association rules mining
论文作者
论文摘要
最近，已经提出了一种用于数据挖掘的根本重要任务的量子算法，联想规则挖掘（ARM）称为简称QARM。值得注意的是，QARM在其经典同行方面实现了ARM的主要任务，即从事务数据库中找到频繁的项目集。在本文中，我们通过IBM量子计算平台实验上在实际量子计算机和量子计算模拟器上实现QARM。首先，我们将QARM的量子电路设计为2 $ \ times $ 2的交易数据库（即，一个涉及两个事务和两个项目的事务数据库），并在四个真实的五Quibit IBM量子计算机以及模拟器上运行它。对于更大的4 $ \ times $ 4交易数据库，它将导致比当前可访问的IBM真实量子设备所能处理的电路更高且深度更高，我们还可以在“ AER \ _Simulator”上构建QARM的量子电路。这两个实验结果都表明，从两个交易数据库中的所有频繁项目集都根据需要成功得出，证明了QARM的正确性和可行性。我们的工作可以用作基准测试，并提供原型，用于在嘈杂的中间尺度量子设备和通用耐断层量子计算机上实现QARM的QARM。
Recently, a quantum algorithm for a fundamentally important task in data mining, association rules mining (ARM), called qARM for short, has been proposed. Notably, qARM achieves significant speedup over its classical counterpart for implementing the main task of ARM, i.e., finding frequent itemsets from a transaction database. In this paper, we experimentally implement qARM on both real quantum computers and a quantum computing simulator via the IBM quantum computing platform. In the first place, we design quantum circuits of qARM for a 2$\times$2 transaction database (i.e., a transaction database involving two transactions and two items), and run it on four real five-qubit IBM quantum computers as well as on the simulator. For a larger 4$\times$4 transaction database which would lead to circuits with more qubits and a higher depth than the currently accessible IBM real quantum devices can handle, we also construct the quantum circuits of qARM and execute them on "aer\_simulator" alone. Both experimental results show that all the frequent itemsets from the two transaction databases are successfully derived as desired, demonstrating the correctness and feasibility of qARM. Our work may serve as a benchmarking, and provide prototypes for implementing qARM for larger transaction databases on both noisy intermediate-scale quantum devices and universal fault-tolerant quantum computers.
