Table 2.
– Runtime profile for the DE approach
| Model type | Dataset | Avg. runtime | Num. iterations | Total runtime |
|---|---|---|---|---|
|
MLP |
Simulated pig | 3.95 min | 2,000 | 131.78 hr |
| Simulated cattle | 4.01 min | 2,000 | 133.67 hr | |
| Real pig | 0.30 min | 10,000 | 49.81 hr | |
|
CNN |
Simulated pig | 2.36 min | 2,000 | 77.67 hr |
| Simulated cattle | 2.73 min | 2,000 | 87.73 hr | |
| Real pig | 0.24 min | 10,000 | 40.15 hr |
MLP, multilayer perceptron; CNN, convolutional neural network; Avg. runtime, average runtime for one DE iteration (each iteration fits two models); Num. iterations, the total number of iterations used in DE; min, minutes; hr, hours.