Significant progress in the classification of Fermi unassociated sources , has led to an increasing number of blazars are being found. The optical spectrum is effectively used to classify the blazars into two groups such as BL Lacs and flat spectrum radio quasars (FSRQs). However, the accurate classification of the blazars without optical spectrum information, i.e., blazars of uncertain type (BCUs), remains a significant challenge. In this paper, we present a principal component analysis (PCA) and machine learning hybrid blazars classification method. The method, based on the data from Fermi LAT 3FGL Catalog, first used the PCA to extract the primary features of the BCUs and then used a machine learning algorithm to further classify the BCUs. Experimental results indicate that the that the use of PCA algorithms significantly improved the classification. More importantly, comparison with the Fermi LAT 4FGL Catalog, which contains the spectral classification of those BCUs in the Fermi-LAT 3FGL Catalog, reveals that the proposed classification method in the study exhibits higher accuracy than currently established methods; specifically, 151 out of 171 BL Lacs and 19 out of 24 FSRQs are correctly classified.

中文翻译：

---

一种对 Fermi-LAT 目录中不确定类型耀变体进行准确分类的混合方法

费米非关联源分类的重大进展导致越来越多的耀变体被发现。光谱有效地用于将耀变体分为两组，例如BL Lacs和平谱射电类星体（FSRQ）。然而，没有光谱信息的耀变体（即不确定类型的耀变体（BCU））的准确分类仍然是一个重大挑战。在本文中，我们提出了一种主成分分析 (PCA) 和机器学习混合 blazars 分类方法。该方法基于 Fermi LAT 3FGL Catalog 的数据，首先使用 PCA 提取 BCU 的主要特征，然后使用机器学习算法对 BCU 进行进一步分类。实验结果表明，PCA算法的使用显着提高了分类效果。更重要的是，与包含 Fermi-LAT 3FGL 目录中那些 BCU 的光谱分类的 Fermi LAT 4FGL 目录进行比较，表明研究中提出的分类方法比目前建立的方法具有更高的准确性；具体来说，171 个 BL Lac 中有 151 个和 24 个 FSRQ 中有 19 个被正确分类。