Browse Source

add chinese version readme.md

tags/v1.2.1
ZhidanLiu 4 years ago
parent
commit
2b81bd635e
6 changed files with 89 additions and 2 deletions
  1. +9
    -2
      README.md
  2. +80
    -0
      README_CN.md
  3. BIN
      docs/differential_privacy_architecture_cn.png
  4. BIN
      docs/differential_privacy_architecture_en.png
  5. BIN
      docs/mindarmour_architecture.png
  6. BIN
      docs/mindarmour_architecture_cn.png

+ 9
- 2
README.md View File

@@ -8,14 +8,21 @@
- [Release Notes](#release-notes)
- [License](#license)

[查看中文](./README_CN.md)

## What is MindArmour

A tool box for MindSpore users to enhance model security and trustworthiness.
A tool box for MindSpore users to enhance model security and trustworthiness and protect privacy data.

MindArmour is designed for adversarial examples, including four submodule: adversarial examples generation, adversarial example detection, model defense and evaluation. The architecture is shown as follow:
MindArmour model security module is designed for adversarial examples, including four submodule: adversarial examples generation, adversarial examples detection, model defense and evaluation. The architecture is shown as follow:

![mindarmour_architecture](docs/mindarmour_architecture.png)

MindArmour differential privacy module Differential-Privacy implements the differential privacy optimizer. Currently, SGD, Momentum and Adam are supported. They are differential privacy optimizers based on the Gaussian mechanism.
This mechanism supports both non-adaptive and adaptive policy. Rényi differential privacy (RDP) and Zero-Concentrated differential privacy(ZDP) are provided to monitor differential privacy budgets. The architecture is shown as follow:

![dp_architecture](docs/differential_privacy_architecture_en.png)

## Setting up MindArmour

### Dependencies


+ 80
- 0
README_CN.md View File

@@ -0,0 +1,80 @@
# MindArmour

- [简介](#简介)
- [开始](#开始)
- [文档](#文档)
- [社区](#社区)
- [贡献](#贡献)
- [版本](#版本)
- [版权](#版权)

[View English](./README.md)

## 简介

MindArmour可用于增强模型的安全可信、保护用户的数据隐私。

模型安全主要针对对抗样本,包含了4个子模块:对抗样本的生成、对抗样本的检测、模型防御、攻防评估。对抗样本的架构图如下:

![mindarmour_architecture](docs/mindarmour_architecture_cn.png)

隐私保护支持差分隐私,包括动态或者非动态的差分隐私SGD、Momentum、Adam优化器,噪声机制支持高斯分布噪声、拉普拉斯分布噪声,差分隐私预算监测包含ZDP、RDP。差分隐私的架构图如下:

![dp_architecture](docs/differential_privacy_architecture_cn.png)


## 开始

### 环境依赖

MindArmour使用MindSpore加速计算,因此请先安装MindSpore,其余依赖包含在`setup.py`里。

### 安装

#### 源码安装

1. 从Gitee下载源码。

```bash
git clone https://gitee.com/mindspore/mindarmour.git
```

2. 编译安装MindArmour。

```bash
$ cd mindarmour
$ python setup.py install
```

#### `Pip`安装

1. 从[MindSpore官网](https://www.mindspore.cn/versions/en)下载whl格式的安装包,然后执行如下命令:

```
pip install mindarmour-{version}-cp37-cp37m-linux_{arch}.whl
```

2. 执行如下命令,如果没有报错`No module named 'mindarmour'`,则说明安装成功。

```bash
python -c 'import mindarmour'
```

## 文档

安装指导、使用教程、API,请参考[用户文档](https://gitee.com/mindspore/docs)。

## 社区

社区问答:[MindSpore Slack](https://join.slack.com/t/mindspore/shared_invite/enQtOTcwMTIxMDI3NjM0LTNkMWM2MzI5NjIyZWU5ZWQ5M2EwMTQ5MWNiYzMxOGM4OWFhZjI4M2E5OGI2YTg3ODU1ODE2Njg1MThiNWI3YmQ)。
## 贡献

欢迎参与社区贡献,详情参考[Contributor Wiki](https://gitee.com/mindspore/mindspore/blob/master/CONTRIBUTING.md)。

## 版本

版本信息参考:[RELEASE](RELEASE.md)。

## 版权

[Apache License 2.0](LICENSE)

BIN
docs/differential_privacy_architecture_cn.png View File

Before After
Width: 1314  |  Height: 690  |  Size: 38 kB

BIN
docs/differential_privacy_architecture_en.png View File

Before After
Width: 1389  |  Height: 700  |  Size: 50 kB

BIN
docs/mindarmour_architecture.png View File

Before After
Width: 699  |  Height: 546  |  Size: 28 kB Width: 645  |  Height: 540  |  Size: 28 kB

BIN
docs/mindarmour_architecture_cn.png View File

Before After
Width: 647  |  Height: 569  |  Size: 18 kB

Loading…
Cancel
Save