Novel methods for rapidly estimating single-photon source (SPS) quality, e.g.
of quantum dots, have been promoted in recent literature to address the
expensive and time-consuming nature of experimental validation via intensity
interferometry. However, the frequent lack of uncertainty discussions and
reproducible details raises concerns about their reliability. This study
investigates one such proposal on eight datasets obtained from an InGaAs/GaAs
epitaxial quantum dot that emits at 1.3 {\mu}m and is excited by an 80 MHz
laser. The study introduces a novel contribution by employing data
augmentation, a machine learning technique, to supplement experimental data
with bootstrapped samples. Analysis of the SPS quality metric, i.e. the
probability of multi-photon emission events, as derived from efficient
histogram fitting of the synthetic samples, reveals significant uncertainty
contributed by stochastic variability in the Poisson processes that describe
detection rates. Ignoring this source of error risks severe overconfidence in
both early quality estimates and claims for state-of-the-art SPS devices.
Additionally, this study finds that standard least-squares fitting is
comparable to the studied counter-proposal, expanding averages show some
promise for early estimation, and reducing background counts improves fitting
accuracy but does not address the Poisson-process variability. Ultimately, data
augmentation demonstrates its value in supplementing physical experiments; its
benefit here is to emphasise the need for a cautious assessment of SPS quality.

使用数据增强等机器学习技术结合物理实验，对单光子源 (SPS) 质量评估方法进行研究，发现忽略探测率泊松过程中的随机变化对 SPS 质量评估会带来严重的过度自信，提出标准最小二乘拟合具有可比性、扩展平均值具有提前预估的优势。降低背景计数可以提高拟合精度，但不能解决泊松过程的可变性问题。

快速确定单光子源的质量挑战

The Challenge of Quickly Determining the Quality of a Single-Photon  Source

While quantum dots are at the forefront of quantum device technology, tuning
multi-dot systems requires a lengthy experimental process as multiple
parameters need to be accurately controlled. This process becomes increasingly
time-consuming and difficult to perform manually as the devices become more
complex and the number of tuning parameters grows. In this work, we present a
crucial step towards automated tuning of quantum dot qubits. We introduce an
algorithm driven by machine learning that uses a small number of coarse-grained
measurements as its input and tunes the quantum dot system into a pre-selected
charge state. We train and test our algorithm on a GaAs double quantum dot
device and we consistently arrive at the desired state or its immediate
neighborhood.

本文介绍了一种基于机器学习的算法，使用少量粗粒度测量值作为输入，将量子点系统调整到预选的电荷状态，为自动调整量子点 qubit 迈出了关键一步。作者还将该算法在 GaAs 双量子点器件上进行了训练和测试，并成功将系统成功调整到预期状态或其相邻区域。