Robustness Verification on 7-class underwater plankton dataset (SilCam) (test)
0Computation Time (s)Neural ODE-based classifier
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Neural ODE-based classifierActivation Function=ReLU, Pixels Attacked=192 / 192, Noise (L∞)=5/2552026.07 | 0 | — | |
| Neural ODE-based classifierActivation Function=ReLU, Pixels Attacked=192 / 192, Noise (L∞)=10/2552026.07 | 0 | — | |
| Neural ODE-based classifierActivation Function=Tanh, Pixels Attacked=192 / 192, Noise (L∞)=5/2552026.07 | 0 | — | |
| Neural ODE-based classifierActivation Function=Tanh, Pixels Attacked=192 / 192, Noise (L∞)=10/2552026.07 | 0 | — | |
| Neural ODE-based classifierActivation Function=Tanh, Pixels Attacked=1 / 192, Noise (L∞)=0.5/2552026.07 | 101.4 | — | |
| Neural ODE-based classifierActivation Function=ReLU, Pixels Attacked=1 / 192, Noise (L∞)=0.5/2552026.07 | 104.2 | — | |
| Neural ODE-based classifierActivation Function=ReLU, Pixels Attacked=10 / 192, Noise (L∞)=0.1/2552026.07 | 104.8 | — | |
| Neural ODE-based classifierActivation Function=Tanh, Pixels Attacked=10 / 192, Noise (L∞)=0.1/2552026.07 | 105.2 | — | |
| Neural ODE-based classifierActivation Function=Tanh, Pixels Attacked=192 / 192, Noise (L∞)=0.05/2552026.07 | 105.2 | — | |
| Neural ODE-based classifierActivation Function=ReLU, Pixels Attacked=192 / 192, Noise (L∞)=0.01/2552026.07 | 108.9 | — | |
| Neural ODE-based classifierActivation Function=Tanh, Pixels Attacked=192 / 192, Noise (L∞)=0.01/2552026.07 | 113.1 | — | |
| Neural ODE-based classifierActivation Function=Tanh, Pixels Attacked=1 / 192, Noise (L∞)=0.1/2552026.07 | 119.2 | — | |
| Neural ODE-based classifierActivation Function=ReLU, Pixels Attacked=1 / 192, Noise (L∞)=0.1/2552026.07 | 128.1 | — | |
| Neural ODE-based classifierActivation Function=ReLU, Pixels Attacked=192 / 192, Noise (L∞)=1/2552026.07 | 7,200 | — | |
| Neural ODE-based classifierActivation Function=ReLU, Pixels Attacked=192 / 192, Noise (L∞)=0.05/2552026.07 | 7,200 | — | |
| Neural ODE-based classifierActivation Function=Tanh, Pixels Attacked=192 / 192, Noise (L∞)=1/2552026.07 | 7,200 | — |