Table 5 Execution time (hours) per 100 epochs on one NVIDIA GPU with a learning rate 1e−4 for the considered scenarios
Model | Execution time, hours |
|---|---|
\({\mathcal L}_{\text {One-class}}^1: -\mathcal {D}_{{\textrm{t}} \hat{\textrm{t}}}\) | 2.78–3.05 |
\({\mathcal L}_{\text {One-class}}^2: -\mathcal {D}_{{\textrm{t}} \hat{\textrm{t}}} + \mathcal {D}_{\textrm{t}}\) | 5.12–5.25 |
\({\mathcal L}_{\text {One-class}}^3: -\mathcal {D}_{{\textrm{t}} \hat{\textrm{t}}} - \beta \mathcal {D}_{{\textrm{x}} \hat{\textrm{x}}}\) | 5.56–5.83 |
\({\mathcal L}_{\text {One-class}}^4: -\mathcal {D}_{\textrm{t} \mathrm{\hat{t}}} + \mathcal {D}_{\textrm{t}} - \beta \mathcal {D}_{{\textrm{x}} \hat{\textrm{x}}} + \beta \mathcal {D}_{\textrm{x}}\) | 11.11–11.39 |