基于自编码神经网络的Single-Pass聚类事件识别算法

针对传统Single-Pass聚类算法存在的缺陷，提出了一种基于自编码神经网络的Single-Pass聚类算法。通过多个深层的隐藏层对原始数据进行降维，以更好地提取出原始数据的特征信息；并通过对边缘文本重计算来降低误检率，提高聚类精度。实验结果表明，该算法相比传统Single-Pass算法具有更高的聚类准确度，解决了聚类结果受数据顺序影响的问题。

The combination of an autoencoder network and Single-Pass clustering for detection and tracking

The traditional Single-Pass clustering algorithm has some deficiencies, such as having relatively low accuracy and requiring complex calculations. Therefore a detection and tracking method based on the combination of an autoencoder network and Single-Pass clustering is proposed in this paper. The original data is refactored by training a neural network with multiple hidden layers, which can better extract the data features. By virtue of establishing a better weighting factor and setting up edge articles, the false detection rate is reduced and the effect of clustering is improved. In addition, the new method overcomes negative effects of the data sequence. The experimental results show that the algorithm is more efficient than the traditional Single-Pass algorithm.