Advances, Systems and Applications
Table 9 Best configuration based on the number of step-ahead prediction sizes
From: VTGAN: hybrid generative adversarial networks for cloud workload prediction
Prediction sizes | Value regression | Trend classification |
|---|---|---|
1-step-ahead | VTGAN (LSTM-based) | VTGAN (LSTM-based) |
| Â | Window size=3 | Window size=3 |
|  | (Fig. 14(a)) | (Fig. 15(a)) |
3-step-ahead | Stacked LSTM | CNN-LSTM |
| Â | Window size=10 | Window size=20 |
| Â | With TIs | With TIs |
|  | (Fig. 14(h)) | (Fig. 15(n)) |
5-step-ahead | Stacked GRU | CNN-LSTM |
| Â | Window size=3 | Window size=3 |
| Â | With TIs | With TIs |
|  | (Fig. 14(c)) | (Fig. 15(c)) |