A Network Data Reinforcement Method Based on the Multiclass Variational Autoencoder

Table 7

The metrics of ADMs on A-Data and raw data.

Model	ADM on reinforced data	ADM on A-Data	SVM on A-Data
Loss convergence during training	0.0010	0.0162	—
Accuracy convergence during training	99.99%	99.77%	—
Accuracy during evaluation	95.29%	99.75%	99.62%
Precision during evaluation	97.86%	99.84%	99.78%
Recall during evaluation	90.90%	99.59%	99.59%
Training time	9.82 s	9.43 s	184.90 s
Evaluation time	2.25 s	2.28 s	6.11 s