MSTP Network Data Traffic Anomaly Optimization Detection Algorithm

The current conventional network data traffic anomaly optimization detection algorithm mainly achieves the extraction of anomaly features by constructing a feature attribute matrix, which leads to poor detection results due to the lack of evaluation of feature importance. In this regard, the MSTP network data traffic anomaly optimization detection algorithm is proposed. By analyzing the mapping relationship between key values and query values, a self-attentive mechanism is constructed to extract network data traffic features. And by calculating the importance change value of feature data before and after adding random noise, feature importance evaluation is realized and anomaly optimization detection model is constructed. In the experiment, the designed model was tested for detection effectiveness. The final results can prove that the algorithm has a high AUC value and a more desirable detection performance when the proposed method is used for anomaly detection of network data traffic.