MCPcopy Index your code
hub / github.com/Trusted-AI/adversarial-robustness-toolbox

github.com/Trusted-AI/adversarial-robustness-toolbox @1.20.1 sqlite

repository ↗ · DeepWiki ↗ · release 1.20.1 ↗
4,583 symbols 21,134 edges 629 files 2,413 documented · 53%
README

Adversarial Robustness Toolbox (ART) v1.20

CodeQL Documentation Status PyPI codecov Code style: black License: MIT PyPI - Python Version slack-img Downloads Downloads CII Best Practices

LF AI & Data

对抗性鲁棒性工具集(ART)是用于机器学习安全性的Python库。ART 由 Linux Foundation AI & Data Foundation (LF AI & Data)。 ART提供的工具可 帮助开发人员和研究人员针对以下方面捍卫和评估机器学习模型和应用程序: 逃逸,数据污染,模型提取和推断的对抗性威胁。ART支持所有流行的机器学习框架 (TensorFlow,Keras,PyTorch,scikit-learn,XGBoost,LightGBM,CatBoost,GPy等),所有数据类型 (图像,表格,音频,视频等)和机器学习任务(分类,物体检测,语音识别, 生成模型,认证等)。

Adversarial Threats

ART for Red and Blue Teams (selection)

学到更多

开始使用 文献资料 贡献
- 安装

该库正在不断开发中。欢迎反馈,错误报告和贡献!

致谢

本材料部分基于国防高级研究计划局(DARPA)支持的工作,合同编号HR001120C0013。 本材料中表达的任何意见,发现和结论或建议均为作者的观点,并不一定反映国防高级研究计划局(DARPA)的观点。

Core symbols most depended-on inside this repo

from_numpy
called by 231
art/attacks/evasion/laser_attack/laser_attack.py
predict
called by 174
art/attacks/extraction/functionally_equivalent_extraction.py
clip
called by 125
art/attacks/evasion/laser_attack/laser_attack.py
_apply_preprocessing
called by 121
art/estimators/estimator.py
check_and_transform_label_format
called by 111
art/utils.py
master_seed
called by 109
tests/utils.py
get_labels_np_array
called by 109
art/utils.py
generate
called by 82
art/attacks/evasion/simba.py

Shape

Method 2,893
Function 1,113
Class 568
Route 9

Languages

Python100%

Modules by API surface

tests/estimators/classification/test_scikitlearn.py96 symbols
conftest.py75 symbols
tests/utils.py64 symbols
art/attacks/evasion/brendel_bethge.py59 symbols
art/estimators/classification/scikitlearn.py55 symbols
art/utils.py54 symbols
art/attacks/attack.py54 symbols
art/attacks/evasion/pixel_threshold.py44 symbols
tests/test_data_generators.py36 symbols
art/estimators/classification/pytorch.py36 symbols
tests/estimators/regression/test_keras_regression.py35 symbols
art/estimators/regression/pytorch.py34 symbols

Dependencies from manifests, versioned

GPy1.13.2 · 1×
Pillow11.3.0 · 1×
black25.1.0 · 1×
catboost1.2.8 · 1×
cma4.2.0 · 1×
ffmpeg-python0.2.0 · 1×
h5py3.14.0 · 1×
keras3.10.0 · 1×
kornia0.8.1 · 1×
librosa0.11.0 · 1×
lief0.16.6 · 1×
lightgbm4.6.0 · 1×

For agents

$ claude mcp add adversarial-robustness-toolbox \
  -- python -m otcore.mcp_server <graph>

⬇ download graph artifact