A supervised machine learning approach using naive Gaussian Bayes classification for shape-sensitive detector pulse discrimination in positron annihilation lifetime spectroscopy (PALS)
A supervised machine learning approach using naive Gaussian Bayes classification for shape-sensitive detector pulse discrimination in positron annihilation lifetime spectroscopy (PALS)
复制标题
DOI：
10.1016/j.nima.2019.162742
复制
发表时间：
2019-12-11
影响因子：
1.4
通讯作者：
Staab, Torsten E. M.
中科院分区：
文献类型：
--
作者：
Petschke, Danny;Staab, Torsten E. M.
获得高质量和无伪影的正电子寿命谱对于深入分析是至关重要的，即正确的寿命谱分解以恢复真实信息。自数字正电子寿命谱仪问世以来，通常通过基于软件的脉冲滤波来实现对探测器脉冲的识别，基于正态分布似然的朴素贝叶斯分类模型，提出了一种新的基于监督机器学习的形状敏感探测器脉冲识别方法。总的来说，朴素贝叶斯方法被广泛应用于许多现实世界的问题，如著名的电子邮件垃圾邮件过滤、文本分类或文档分类。他们的算法相对容易实现，而且，与在高维数据集(即检测器脉冲)上训练和预测的更复杂的方法相比，执行速度极快。在这项研究中，我们表明，少于20个标记的训练脉冲的显著低数量足以获得与应用物理滤波类似的结果。因此，我们的方法代表了一种潜在的替代方案。
The acquisition of high-quality and non-artefact afflicted positron lifetime spectra is crucial for a profound analysis, i.e. the correct lifetime spectra decomposition for retrieving the true information. Since the introduction of digital positron lifetime spectrometers, this is generally realized by applying detector pulse discrimination with the help of software-based pulse filtering regarding area and/or shape of the detector pulses.Here, we present a novel approach for shape-sensitive detector pulse discrimination applying supervised machine learning (ML) based on a naive Bayes classification model using a normally distributed likelihood. In general, naive Bayes methods find wide application for many real-world problems such as famously applied for email spam filtering, text categorization or document classification. Their algorithms are relatively simple to implement and, moreover, perform extremely fast compared to more sophisticated methods in training and predicting on high-dimensional datasets, i.e. detector pulses. In this study we show that a remarkable low number of less than 20 labelled training pulses is sufficient to achieve comparable results as of applying physically filtering. Hence, our approach represents a potential alternative.