Small sample malicious traffic classification method and system based on deep migration

The invention belongs to the technical field of network traffic detection, and particularly relates to a small sample malicious traffic classification method and system based on deep migration, and the method comprises the steps: firstly carrying out the preprocessing of collected small sample malicious traffic data; small sample malicious traffic data distribution is adjusted through data mapping, so that the data distribution is adjusted to be source data set data distribution meeting a preset similarity condition; pre-training the traffic classification model by using the source data set, taking small sample malicious traffic data as a target data set, obtaining recommended frozen network parameters by using a dichotomy task between the source data set and the target data set, adjusting the parameters of the traffic classification model, and finely adjusting the traffic classification model through the target data set; and performing classification detection on malicious traffic generated in the target net