基于对无监督学习算法特点即及光纤传感事件信号的时空数据特点的研究初步决定通过使用以下常用无监督学习算法对光纤传感事件进行分类：n①聚类n1基于划分聚类的K-means算法、n2n3n4n5n②降维：n1n2n3n4n同时深入对无监督学习的深度学习尝试确定型的自编码方法及其改进算法和概率型的受限波尔兹曼机及其改进算法两类深度算法在光纤事件分类上的应用。nn其次在无监督的基础上应用半监督学习算法研究其应用效果。初步尝试的算法为：n最后初步调试了无监督K-means算法nn完善上述内容。

Based on the characteristics of unsupervised learning algorithms and the spatiotemporal data features of fiber optic sensing events, we have preliminarily decided to use the following commonly used unsupervised learning algorithms for fiber optic sensing event classification:

① Clustering (1) K-means algorithm based on partition clustering (2) (3) (4) (5)

② Dimensionality reduction: (1) (2) (3) (4)

Meanwhile, we will delve into the deep learning of unsupervised learning, trying to determine the application of two types of deep algorithms, namely, the deterministic autoencoder method and its improved algorithm and the probabilistic restricted Boltzmann machine and its improved algorithm in fiber optic event classification.

Secondly, we will apply semi-supervised learning algorithms based on unsupervised learning and study their application effects. The algorithms we have preliminarily tried are:

Finally, we have preliminarily debugged the unsupervised K-means algorithm.

We will continue to improve the above contents.