量子机器学习

人工智能和量子物理是上世纪发展起来的两个截然不同但又影响深远的学科.近年来,它们在数据科学方面的结合引起了学术界的高度关注,形成了量子机器学习这个新兴领域.利用量子态的叠加性,量子机器学习有望通过量子并行解决目前机器学习中数据量大,训练过程慢的困难,并有望从量子物理的角度提出新的学习模型.目前该领域的研究还处于探索阶段,涵盖内容虽然广泛,但还缺乏系统的梳理.本文将从数据和算法角度总结量子机器学习与经典机器学习的不同,以及其中涉及的关键加速技巧,针对数据结构(数字型、模拟型),计算技巧(相位估计、Grover搜索、内积计算),基础算法(求解线性系统、主成分分析、梯度算法),学习模型(支持向量机、近邻法、感知器、玻尔兹曼机)等4个方面对现有研究成果进行综述,并建议一些可能的研究方向,供本领域的研究人员参考.

Quantum machine learning

Artificial intelligence and quantum physics are two most influential disciplines developed in the last century. Recent years, their marriage in data science attracted much attention, forming the frontier field of quantum machine learning. Exploiting quantum superposition, quantum machine learning provides the hope for resolving difficulties in big data and training process, and new learning models illuminated by quantum physics. So far, the field is still in its infancy. Although many problems had been explored, a systematic theory is still lacking. This review will summarize the major differences in data structure and algorithms between classical and quantum machine learning, as well as the key acceleration techniques. Several topics will be covered, including the data structure (digital and analog), computation skills (phase estimation, Grover search and inner product), fundamental algorithms (linear equation, principle component analysis and gradient algorithms) and several learning models (supporting vector machine, nearest-neighbor method, perceptron networks and Boltzmann machine). Finally, we propose several promising directions in this field.