|
pulsatrix 1.1.0
|
ExplanationSensitivity()'s result. More...
#include <explanation_robustness.hpp>
Public Attributes | |
| double | max_sensitivity = 0.0 |
| The largest and the mean relative change of the explanation over the kept samples; 0 when none were kept. | |
| double | avg_sensitivity = 0.0 |
| double | local_lipschitz = 0.0 |
| The largest change of the explanation per unit change of the input. | |
| std::optional< double > | adversarial_sensitivity |
| The relative change at FGSM's point; empty without a classifier, or when every halving still changed the prediction. | |
| double | adversarial_epsilon = 0.0 |
| FGSM's L∞ step at the adversarial point. | |
| int64_t | samples = 0 |
| The samples drawn, and those that kept the predicted class. | |
| int64_t | kept = 0 |
| double | radius = 0.0 |
| The L∞ radius the samples were drawn in, in the input's units. | |
ExplanationSensitivity()'s result.
| double pulsatrix::SensitivityReport::adversarial_epsilon = 0.0 |
FGSM's L∞ step at the adversarial point.
| std::optional<double> pulsatrix::SensitivityReport::adversarial_sensitivity |
The relative change at FGSM's point; empty without a classifier, or when every halving still changed the prediction.
| double pulsatrix::SensitivityReport::avg_sensitivity = 0.0 |
| int64_t pulsatrix::SensitivityReport::kept = 0 |
| double pulsatrix::SensitivityReport::local_lipschitz = 0.0 |
The largest change of the explanation per unit change of the input.
| double pulsatrix::SensitivityReport::max_sensitivity = 0.0 |
The largest and the mean relative change of the explanation over the kept samples; 0 when none were kept.
| double pulsatrix::SensitivityReport::radius = 0.0 |
The L∞ radius the samples were drawn in, in the input's units.
| int64_t pulsatrix::SensitivityReport::samples = 0 |
The samples drawn, and those that kept the predicted class.