Network Fault Classification Techniques Based on Transfer Learning

Aiming at the scarcity of fault samples in data network fault classification, this paper proposes a transfer learning method based on parameter sharing scheme. Using the network propagation characteristics of fault information of network elements, the known knowledge of source domain model is transferred to the target domain with a few samples by transferring the source domain model parameters to the target model for training, which improves the generalization ability of the model to the network topology, and solves the problems of scarcity of labeled samples and dynamic network topology. By comparing the classification performance of the transfer learning model under different network model and parameters, the network model was gradually tuned. The experiments show that the optimized model's fault classification accuracy rate reaches 99.4%.