A Deep Learning-based Model for Anomaly Data Detection in Cloud Failure Scenarios

The rapid development of cloud platform makes the traffic data show exponential growth. In order to ensure the stable operation of the business required by users, it is necessary to quickly detect the anomalous data in cloud platform failure scenarios. Considering the seriousness of the current network security situation as well as the problems of data clutter, data imbalance and incomplete feature learning of traditional network anomaly detection models, this paper proposes a deep learning data anomaly detection model based on CNN-BiLSTM-Attention and introduces the ADASYN algorithm to balance the number of samples. Experiments show that the model achieves 97% accuracy in Yahoo dataset and can effectively detect data anomalies in cloud platform.