A Deep Learning Algorithm Using Feature Engineering to Adjust Attention Mechanisms and Neural Network for Cloud Security Detection

Abstract

AbstractCloud computing realizes the intensive management of resources and improves the production efficiency, but it also inevitably brings security problems. Cloud security detection technology can be understood as immune cells in the human body, which can protect against all kinds of viruses invading the server. In this article, we transform the virus activity information into time series data, and study how to classify the virus through deep learning algorithm. In our system, we roughly divide the virus activity information into six categories, and calculate the importance of these six types of behavior to virus recognition through feature engineering. Then, we integrate the calculated parameters into the construction of attention mechanism layer and neural network embedding layer to propose a new algorithm based on deep learning architecture. Finally, we verify the accuracy of the new algorithm in virus classification through open source dataset. We compare the performance of this model with logistic regression, support vector machine, random forest model and convolutional neural network. The experimental results show that our model has certain advantages in F1, and improves the performance index by nearly 4%.KeywordsDeep learningFeature engineeringAttention mechanismsCloud security detection