adversarial-robustness-toolbox - Adversarial Robustness Toolbox (ART) - Python Libr

项目说明

Ad Robustness Toolbox (ART) v1.20

<p align="center">
<img src=" width="467" title="ART logo">
</p>
<br />

!CodeQL
![Documentation Status]()
![PyPI]()
![codecov]()
![Code style: black]()
![License: MIT]()
![PyPI - Python ]()
![slack-img]()
![Downloads]()
![Downloads]()
![CII Best Practices]()

中文README请按此处

<di="center">
<picture>
<source media="(prefers-color-scheme: dark)" srcset="docs/images/lfaidata-project-badge-graduate-color_dark.png" width="400" title="LF AI & Data">
<source media="(prefers-color-scheme: light)" srcset="docs/images/lfaidata-project-badge-graduate-color.png" width="400" title="LF AI & Data">
<img alt="Fallback image description" src="default-image.png" width="400">
</picture>
</div>
<br />

Ad Robustness Toolbox (ART) is a Python library for Machine Learning Security. ART is hosted by the
Linux Foundation AI & Data Foundation (LF AI & Data). ART provides tools that enable
de and researchers to defend and e Machine Learning models and applications against the
ad threats of E, Poisoning, Extraction, and Inference. ART supports all popular machine learning frameworks
(TensorFlow, Keras, PyTorch, scikit-learn, XGBoost, LightGBM, CatBoost, GPy, etc.), all data types
(images, tables, audio, video, etc.) and machine learning tasks (classification, object detection, speech recognition,
generation, certification, etc.).

Ad Threats

<di="center">
<picture>
<source media="(prefers-color-scheme: dark)" srcset="docs/images/ader_dark.png" width="400 title="ART Threats">
<source media="(prefers-color-scheme: light)" srcset="docs/images/ader.png" width="400 title="ART Threats">
<img alt="Fallback image description" src="default-image.png" width="400">
</picture>
</div>

<p align="center">
<img src="docs/images/ad.png?raw=true" width="400" title="ART Matrix">
</p>
<br />

ART for Red and Blue Teams (selection)

<di="center">
<picture>
<source media="(prefers-color-scheme: dark)" srcset="docs/images/white_hat_blue_red_dark.png" width="800 title="ART Red and Blue Teams">
<source media="(prefers-color-scheme: light)" srcset="docs/images/white_hat_blue_red.png" width="800 title="ART Red and Blue Teams">
<img alt="Fallback image description" src="default-image.png" width="800">
</picture>
</div>
<br />

Learn more

| [Get Started][get-started] | [Documentation][documentation] | [Contributing][contributing] |
|-------------------------------------|-------------------------------|-----------------------------------|
| - [Installation][installation]<br>- Examples<br>- Notebooks | - [Attacks][attacks]<br>- [Defences][defences]<br>- [Estimators][estimators]<br>- [Metrics][metrics]<br>- Technical Documentation | - Slack, In<br>- Contributing<br>- [Roadmap][roadmap]<br>- [Citing][citing] |

[get-started]:
[attacks]:
[defences]:
[estimators]:
[metrics]:
[contributing]:
[documentation]:
[installation]:
[roadmap]:
[citing]:

The library is under continuous de. Feedback, bug reports and contributions are very welcome!

Acknowledgment

This material is partially based upon work supported by the Defense Ad Research Projects Agency (DARPA) under
Contract No. HR001120C0013. Any opinions, findings and conclusions or recommendations expressed in this material are
those of the author(s) and do not necessarily reflect the views of the Defense Ad Research Projects Agency (DARPA).

项目信息

开发语言：Python

Star 数：6254

技术要点

该项目属于红蓝攻防方向的开源资源，代码结构清晰，适合安全研究人员学习参考。项目围绕Adversarial Robustness Toolbox (ART) - Python Library for Machine Learning Security - Evasion, Poisoning, Extraction, Inference - Red and Blue Teams展开，提供了可运行的实现，对理解相关安全机制、工具原理与攻防技术具有较高的研究价值。

从技术实现角度看，此类项目的核心在于对底层原理的深入理解与工程化落地。开发者在研究过程中可以结合官方文档与社区讨论，梳理项目的初始化流程、核心模块职责与数据流转，逐步掌握其中的技术细节与设计思路。

使用与学习建议

1. 下载源码压缩包，解压到本地环境。

2. 根据项目类型准备对应的运行环境与依赖。

3. 按照 README 中的步骤完成配置与部署。

4. 在隔离的测试环境中运行验证，避免影响生产系统。

免责声明

本资源来源于开源社区，仅供安全研究、技术学习与合法授权范围内的攻防演练使用，请勿用于任何非法或侵害他人权益的用途。使用本项目产生的一切后果由使用者自行承担。