Network attack detection method and device based on deep k-nearest neighbor

Abstract

The invention discloses a network attack detection method and device based on deep k-nearest neighbor. The method comprises the steps: constructing a deep network model for network attack detection, training the deep network model through employing training sample data, obtaining a trained deep network model, inputting the training sample data into the trained deep network model, and obtaining a network attack detection result. Obtaining the output of training sample data in each layer of the deep network model, inputting correction sample data into the trained deep network model, calculating to obtain the sum of distance features corresponding to the correction sample data to form a data set, then inputting to-be-detected data into the trained deep network model, and combining DkNN analysis to obtain a to-be-detected data set; and obtaining a prediction result. The technical scheme of the invention has good robustness, and more accurate credibility of the prediction classification result is obtained.

Description

technical field [0001] The invention belongs to the field of network attack defense and deep learning, and specifically relates to a network attack detection method and device based on deep k-nearest neighbors, which are used for intrusion detection in vulnerable network environments. Background technique [0002] With the development and maturity of the Internet, the Internet has gradually become a highly shared and free environment. The networking and intelligence of all walks of life have begun to flourish. But at the same time, various malicious network activities and cyber crimes occur frequently, and the information security of network users is seriously threatened. [0003] In order to solve the above problems, the intrusion detection system, as a representative network defense technology, has become popular again in recent years. The intrusion detection system obtains the characteristics of the attack activity by analyzing the attack activity after the event, and w...

AI Technical Summary

Problems solved by technology

[0007] The technical problem to be solved by the present invention is: the reliability of the prediction of the intrusion detection output model based on the traditional neural network is generally high, and the robustness of the model is low, and the performance is poor when faced with adversarial examples

Images