Construction of a Computer Data Security Detection Model based on Artificial Intelligence

In the information society, the network has penetrated into all aspects of people's life, work and study, providing great convenience for people. Hackers attack through key devices in the Internet in a very complex way, and the protection of terminal equipment is often difficult to play an effective role due to the limitations of its operating environment, which in turn restricts the subsequent prevention and optimisation of network security. Network traffic anomaly detection methods based on machine learning have gradually been emphasized. This paper introduces a new effective method based on deep learning. Firstly, the service information is transformed into feature vectors with better representativeness through network service data transformation; secondly, the extracted feature vectors are used to model the abnormal behaviors, and the models are trained and validated; finally, the new services are classified into normal and abnormal services. In this paper, a model of CNN-BiGRU (Convolutional Neural Network with Bidirectional Gated Recurrent Unit) combined with an attention mechanism is used to extract local features and time series dependent features. The experimental results show that the model exhibits different performances under different datasets, training set proportions, time steps and optimizer conditions. The experimental data includes a variety of data security related datasets. The results show that when the time step is set to 200, the model achieves an accuracy of 93.0%, a precision of 92.0%, a recall of 94.0%, and an F1 score of 93.0. The study provides an effective technical solution in the field of network data security detection, and provides reliable support and guidance for network security defense.