Adversarial examples in neural networks
GPTKB entity
Statements (49)
| Predicate | Object |
|---|---|
| gptkbp:instanceOf |
research
|
| gptkbp:application |
security testing
robustness evaluation |
| gptkbp:challenge |
certifying robustness
defending against adversarial attacks detecting adversarial examples |
| gptkbp:concerns |
AI safety
model vulnerability trustworthiness of AI systems |
| gptkbp:defenseMechanism |
adversarial training
robust optimization input preprocessing |
| gptkbp:defines |
Inputs to machine learning models that are intentionally designed to cause the model to make a mistake.
|
| gptkbp:field |
gptkb:artificial_intelligence
gptkb:machine_learning |
| gptkbp:firstDescribed |
2013
Szegedy et al. |
| https://www.w3.org/2000/01/rdf-schema#label |
Adversarial examples in neural networks
|
| gptkbp:impact |
can cause misclassification
can reduce model accuracy |
| gptkbp:method |
adding small perturbations to input data
|
| gptkbp:notableBattle |
Carlini & Wagner attack
DeepFool Fast Gradient Sign Method (FGSM) Jacobian-based Saliency Map Attack (JSMA) Projected Gradient Descent (PGD) |
| gptkbp:notableContributor |
gptkb:Ian_Goodfellow
gptkb:Christian_Szegedy gptkb:Dawn_Song gptkb:Alexey_Kurakin Nicholas Carlini Nicolas Papernot |
| gptkbp:notablePublication |
Explaining and Harnessing Adversarial Examples (Szegedy et al., 2013)
Towards Evaluating the Robustness of Neural Networks (Carlini & Wagner, 2017) Adversarial Machine Learning at Scale (Kurakin et al., 2016) Intriguing properties of neural networks (Szegedy et al., 2013) |
| gptkbp:relatedConcept |
robustness
explainability transferability black-box attack gradient masking white-box attack |
| gptkbp:relatedTo |
deep learning
neural networks |
| gptkbp:trainer |
gptkb:CIFAR-10
gptkb:ImageNet gptkb:MNIST |
| gptkbp:bfsParent |
gptkb:Christian_Szegedy
|
| gptkbp:bfsLayer |
7
|