Encrypted traffic classification method based on incremental learning

Abstract

The invention provides a traffic encryption method based on incremental learning, which relates to the field of network classification traffic, and comprises the following steps of: 1, preprocessing original traffic, and converting to generate a three-channel RGB (Red, Green, Blue) image data set; step 2, constructing a deep residual network for the small-size traffic picture; step 3, inputting partial class samples into a deep residual network for training to realize encrypted traffic classification; 4, updating the network model and parameters, and training a loss function based on category balance to minimize the loss function; 5, screening representative old samples in the memory, and retaining old sample learning experience; and 6, repeating the step 3, the step 4 and the step 5. According to the method, incremental learning can be effectively realized from dynamically increasing encrypted traffic data, the classification accuracy is improved, and meanwhile, the model is prevented from inclining a new sample.

Description

technical field [0001] The invention relates to the field of network traffic classification, in particular to an encrypted traffic classification method based on incremental learning. Background technique [0002] With the widespread popularization and dissemination of Internet applications, a large amount of network traffic is generated when people access the Internet. Among them, encrypted traffic is widely used in the network, and its share of the total traffic is increasing year by year. Therefore, the encrypted traffic classification algorithm has gradually attracted extensive attention in academia and industry, and the correct identification and classification of encrypted traffic has become an important topic in network security and network management. On the one hand, traffic operators can use encrypted traffic classification to analyze the proportion of traffic occupied by different network applications, rationally plan networks, improve network management, and imp...

AI Technical Summary

Problems solved by technology

[0004] In the context of the continuous growth of encrypted traffic data, the time complexity and space complexity of these technologies continue to increase with the expansion of data volume

Although the above technologies have achieved good classification results on fixed public datasets, when the amount of data reaches a certain scale, the learning rate of ordinary encrypted traffic classification models often cannot catch up with the rate of data update, which can no longer meet the actual needs

Images