Stealthy Adversarial Attacks on Stochastic Multi-Armed Bandits
Fuente:
arXiv
Guardado en:
| Autores principales: | , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2024
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866913238434709504 |
|---|---|
| author | Wang, Zhiwei Wang, Huazheng Wang, Hongning |
| author_facet | Wang, Zhiwei Wang, Huazheng Wang, Hongning |
| contents | Adversarial attacks against stochastic multi-armed bandit (MAB) algorithms have been extensively studied in the literature. In this work, we focus on reward poisoning attacks and find most existing attacks can be easily detected by our proposed detection method based on the test of homogeneity, due to their aggressive nature in reward manipulations. This motivates us to study the notion of stealthy attack against stochastic MABs and investigate the resulting attackability. Our analysis shows that against two popularly employed MAB algorithms, UCB1 and $ε$-greedy, the success of a stealthy attack depends on the environmental conditions and the realized reward of the arm pulled in the first round. We also analyze the situation for general MAB algorithms equipped with our attack detection method and find that it is possible to have a stealthy attack that almost always succeeds. This brings new insights into the security risks of MAB algorithms. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2402_13487 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Stealthy Adversarial Attacks on Stochastic Multi-Armed Bandits Wang, Zhiwei Wang, Huazheng Wang, Hongning Machine Learning Cryptography and Security Adversarial attacks against stochastic multi-armed bandit (MAB) algorithms have been extensively studied in the literature. In this work, we focus on reward poisoning attacks and find most existing attacks can be easily detected by our proposed detection method based on the test of homogeneity, due to their aggressive nature in reward manipulations. This motivates us to study the notion of stealthy attack against stochastic MABs and investigate the resulting attackability. Our analysis shows that against two popularly employed MAB algorithms, UCB1 and $ε$-greedy, the success of a stealthy attack depends on the environmental conditions and the realized reward of the arm pulled in the first round. We also analyze the situation for general MAB algorithms equipped with our attack detection method and find that it is possible to have a stealthy attack that almost always succeeds. This brings new insights into the security risks of MAB algorithms. |
| title | Stealthy Adversarial Attacks on Stochastic Multi-Armed Bandits |
| topic | Machine Learning Cryptography and Security |
| url | https://arxiv.org/abs/2402.13487 |