Data reconstruction method based on 1D packet convolutional neural network

Abstract

The invention discloses a data reconstruction method based on a 1D packet convolutional neural network. The method comprises the steps of data grouping, model construction, training optimization and data reconstruction. The method specifically comprises the following steps: calculating the correlation between data features and arranging the data features in a descending order, then grouping the data according to the correlation, inputting the data into a grouping convolutional neural network for grouping operation, and outputting reconstructed features through global convolution operation of a full connection layer and feature splicing of a splicing layer to realize feature reconstruction of any dimension. The dimension of the obtained reconstructed feature is reduced, and the space complexity and the time complexity of the model are reduced, so that the time can be reduced, and the memory resource occupation can be reduced. According to the method, the correlation between the features is utilized during grouping, and the correlation between the reconstructed features is improved; by grouping the data, the dimension of the reconstruction feature can be controlled, the data dimension reduction is realized, the operation process of deep learning is simplified, and the operation efficiency of the model is also improved.

Description

technical field [0001] The present invention relates to the field of network security big data analysis and modeling, in particular to a data reconstruction method based on 1D group convolution neural network. Background technique [0002] There are various cyber attacks in cyberspace, such as malicious codes, phishing emails and websites, traffic attacks, exploits, etc. These attacks will not only cause huge economic losses, but even threaten national security and social stability. Detection of cyber threats is necessary. During the detection process, a large amount of network data needs to be collected, such as malware, phishing emails, network traffic, system logs, etc. It is difficult to achieve good results by building traditional machine learning models to analyze these data. With the continuous development of deep learning and artificial intelligence computing, and the successful application of deep learning technology in computer vision, natural language processing,...

AI Technical Summary

Problems solved by technology

When the 2D-CNN model is used to classify network security data, it is first necessary to convert the data into an image format and then process it. In the process of processing, there are defects such as complex calculation process and large amount of calculation.

Images