In the modern world, facial identification is an extremely important task in which many applications rely on high performing algorithms to detect faces efficiently. Whilst classical methods of SVM and k-NN commonly used may perform to a good standard, they are often highly complex and take substantial computing power to run effectively. With the rise of quantum computing boasting large speedups without sacrificing large amounts of much needed performance, we aim to explore the benefits that quantum machine learning techniques can bring when specifically targeted towards facial identification applications. In the following work, we explore a quantum scheme which uses fidelity estimations of feature vectors in order to determine the classification result. Here, we are able to achieve exponential speedups by utilizing the principles of quantum computing without sacrificing large proportions of performance in terms of classification accuracy. We also propose limitations of the work and where some future efforts should be placed in order to produce robust quantum algorithms that can perform to the same standard as classical methods whilst utilizing the speedup performance gains.

中文翻译：

---

使用量子机器学习技术构建人脸识别系统

在现代世界中，人脸识别是一项极其重要的任务，其中许多应用程序依赖高性能算法来有效地检测人脸。虽然常用的 SVM 和 k-NN 的经典方法可以达到很好的标准，但它们通常非常复杂，需要大量的计算能力才能有效运行。随着量子计算的兴起，在不牺牲大量急需的性能的情况下拥有大幅加速，我们的目标是探索量子机器学习技术在专门针对面部识别应用时可以带来的好处。在接下来的工作中，我们探索了一种量子方案，该方案使用特征向量的保真度估计来确定分类结果。这里，我们能够通过利用量子计算原理实现指数加速，而不会牺牲分类精度方面的大部分性能。我们还提出了工作的局限性，以及未来应该在哪些方面做出一些努力，以产生稳健的量子算法，这些算法可以在利用加速性能提升的同时达到与经典方法相同的标准。