Traditional machine learning (ML) an‎d deep learning (DL) based traffic detection an‎d classification methods have low accuracy challenges based on the lack of proper understan‎ding of network traffic characteristics. Traditional machine learning models used in traffic classification assume that training data an‎d test data have identical an‎d independent distributions. However, this assumption may be violated in practical traffic classification due to changes in traffic characteristics, such that models trained with existing data will be ineffective in classifying new traffic. In order to fill these challenges, deep transfer learning networks have recently received much attention. The transfer learning model performs the classification task without considering the limiting assumptions. Transfer learning uses labeled data from the target domain to assess the availability of data in the source domain, with the aim of transferring valuable knowledge from the source domain to the target task. After that, valuable auxiliary data is extracted from the source data an‎d combined with the labeled data in the target domain to train the traffic classifier. Transfer learning helps the learning task in the target domain by transferring useful knowledge from the source domain to the target domain. Therefore, design methods for traffic classification based on transfer learning can be more accurate an‎d realistic. In this research, the topic of Design a Network Traffic Classification Method Based on Deep Transfer Learning will be discussed. Finally, some ideas for improving the existing methods for future research will be presented.

12/1/2025 12:00:00 AM

88938

1404/09/11

Design a Network Traffic Classification Method Based on Deep Transfer Learning

Network Security , Traffic Classification , Deep Learning , Transfer Learning , Source Domain , Target Domain