A Data Anomaly Recognition Method for Mobile Communication Networks Based on Sequential Convolutional Neural Networks
In mobile communication networks,there is usually much more normal data than abnormal data,resulting in serious imbalance problems in the dataset and reducing the accuracy of data anomaly recognition.To overcome this limitation,this study proposes a mobile communication network data anomaly recognition method based on spatial convolutional neural network.Using K-means clustering algorithm to cluster mobile communication network data,in order to reduce noise impact,improve anomaly detection accuracy and computational efficiency,and calculate similarity through Euclidean distance to optimize clustering centers until the sum of squared errors is minimized.Constructing a sequence convolutional neural network model,utilizing its advantages of efficiently extracting local features of sequence data,capturing temporal dependencies,and reducing feature dimensions,to identify anomalies in mobile communication network data.Through convolution,pooling,and fully connected layer processing,combined with Softmax activation function,accurate classification is achieved.The experimental results show that the accuracy of the data anomaly recognition method in the article is high,and the abnormal data results are basically consistent with the actual results,with a maximum recognition error of only 4.
sequential convolutional neural networkmobile communication networkdata anomaly identificationSoftmax activation function
万方数据
维普
