One attack's result.
More...
#include <adversarial.hpp>
|
| Tensor | adversarial |
| | The adversarial inputs, the input's shape, on the input's device.
|
| |
| std::vector< int64_t > | clean_predicted |
| | Each input's predicted class before and after the attack.
|
| |
| std::vector< int64_t > | adversarial_predicted |
| |
| std::vector< bool > | success |
| | Untargeted: the adversarial input isn't classified as its label. Targeted: it is classified as its target.
|
| |
| std::vector< double > | clean_loss |
| | The cross-entropy of each label (the target's, when targeted) before and after.
|
| |
| std::vector< double > | adversarial_loss |
| |
| std::vector< double > | perturbation |
| | Each perturbation's size in the attack's norm: at most epsilon, up to rounding.
|
| |
| double | success_rate = 0.0 |
| | The fraction of inputs the attack succeeded on.
|
| |
| double | robust_accuracy = 0.0 |
| | Untargeted: the fraction classified correctly after the attack.
|
| |
| double | clean_accuracy = 0.0 |
| | Untargeted: the fraction classified correctly before it.
|
| |
| bool | targeted = false |
| |
◆ adversarial
| Tensor pulsatrix::AttackResult::adversarial |
The adversarial inputs, the input's shape, on the input's device.
◆ adversarial_loss
| std::vector<double> pulsatrix::AttackResult::adversarial_loss |
◆ adversarial_predicted
| std::vector<int64_t> pulsatrix::AttackResult::adversarial_predicted |
◆ clean_accuracy
| double pulsatrix::AttackResult::clean_accuracy = 0.0 |
Untargeted: the fraction classified correctly before it.
◆ clean_loss
| std::vector<double> pulsatrix::AttackResult::clean_loss |
The cross-entropy of each label (the target's, when targeted) before and after.
◆ clean_predicted
| std::vector<int64_t> pulsatrix::AttackResult::clean_predicted |
Each input's predicted class before and after the attack.
◆ perturbation
| std::vector<double> pulsatrix::AttackResult::perturbation |
Each perturbation's size in the attack's norm: at most epsilon, up to rounding.
◆ robust_accuracy
| double pulsatrix::AttackResult::robust_accuracy = 0.0 |
Untargeted: the fraction classified correctly after the attack.
◆ success
| std::vector<bool> pulsatrix::AttackResult::success |
Untargeted: the adversarial input isn't classified as its label. Targeted: it is classified as its target.
◆ success_rate
| double pulsatrix::AttackResult::success_rate = 0.0 |
The fraction of inputs the attack succeeded on.
◆ targeted
| bool pulsatrix::AttackResult::targeted = false |
The documentation for this struct was generated from the following file: