Table 1 Methods for numerical prediction and improving prediction performance.
From: Core network traffic prediction based on vertical federated learning and split learning
Index | Training methods | Optimization objectives | Data feature | Key technology |
|---|---|---|---|---|
Ref.[9] | Centralized | Accuracy | Isomorphism | ARIMA |
Ref.[10] | Centralized | Accuracy | Isomorphism | ARIMA |
Ref.[11] | Centralized | Accuracy | Isomorphism | ARIMA and SVR |
Ref.[15] | Centralized | Accuracy | Isomorphism | LSTM and Lasso |
Ref.[16] | Centralized | Accuracy | Isomorphism | Embedding technique and LSTM |
Ref.[17] | Distributed | Accuracy | Isomorphism | Federated learning |
Ref.[18] | Centralized | Accuracy | Heterogeneity | Metapath and LSTM |
Ref.[19] | Centralized | Privacy and accuracy | Isomorphism | Graph convolutional neural network |
Ref.[20] | Centralized | Accuracy | Heterogeneity | Locality-sensitive hashing and LSTM |
Ref.[21] | Centralized | Accuracy | Isomorphism | Graph convolution network |
Proposed | Distributed | Accuracy | Heterogeneity | Vertical federated learning and split learning |
