Advances, Systems and Applications
Table 6 Comparison of classification results
From: VTGAN: hybrid generative adversarial networks for cloud workload prediction
Model | Window size | Training epochs | Time (Sec.) | Precision | Recall | \(\varvec{F_1 score}\) |
|---|---|---|---|---|---|---|
ARIMA | 3 | Â | 35.2 | 0.908 | 0.8681 | 0.8876 |
SVR | 3 | Â | 18.7 | 0.9339 | 0.895 | 0.914 |
VTGAN (LSTM-based) | 3 | 3000 | 74.4 | 0.966±0.003 | 0.900±0.003 | 0.932±0.002 |
CNN-LSTM | 15 | 596 | 80.7 | 0.929±0.007 | 0.890±0.024 | 0.909±0.015 |
Stacked LSTM | 5 | 364 | 55.2 | 0.881±0.008 | 0.864 | 0.873±0.004 |
VTGAN (GRU-based) | 3 | 3000 | 68.7 | 0.954±0.009 | 0.893±0.006 | 0.922±0.007 |
CNN-GRU | 10 | 232 | 35.6 | 0.915±0.02 | 0.853±0.01 | 0.883±0.014 |
Stacked GRU | 5 | 357 | 52.2 | 0.947±0.0002 | 0.900±0.003 | 0.923±0.002 |