Table 1 Summary of GAN-based methods for generating adversarial examples.
From: GEAAD: generating evasive adversarial attacks against android malware defense
Year | Refs. | Model features | Attack method | Evasion capacity (%) | Evasion against |
|---|---|---|---|---|---|
2021 | Generator, attacker and discriminator base model | Semi (Black/White) | 99 | Android malware detector | |
2021 | Generative adversarial network (GAN) to refine generation of adversarial examples | Semi (Black/White) | 98 | Android malware detector | |
2020 | Adversarial-example attack method based on bi-objective GAN | Black box | 95.02 | Android malware detector and firewall | |
2020 | Generate adversarial examples by using any random noise | Semi (Black/White) | 99 | Android malware detector | |
2020 | The intrinsic non-linear structure to generate adversarial examples | White box | 90 | Android malware detector | |
2019 | Dynamically train distilled model with query information | Semi (Black/White) | 92.76 | Android malware detector | |
2019 | Produces query-specific perturbation for query images to form adversarial queries | White box | 100 | Android malware detector | |
2017 | Focuses on the functionality of the substitute detector | Black box | 95.64 | Android malware detector |
