[论文解读] Diffprivlib: The IBM Differential Privacy Library

Naoise Holohan, Stefano Braghin+2 more|arXiv (Cornell University)|Jul 4, 2019

Privacy-Preserving Technologies in Data参考文献 7被引用 66

一句话总结

This paper introduces Diffprivlib, an open-source Python library that provides differential privacy mechanisms, DP-enabled machine learning models, and data analysis tools, with tight integration to NumPy and scikit-learn.

ABSTRACT

Since its conception in 2006, differential privacy has emerged as the de-facto standard in data privacy, owing to its robust mathematical guarantees, generalised applicability and rich body of literature. Over the years, researchers have studied differential privacy and its applicability to an ever-widening field of topics. Mechanisms have been created to optimise the process of achieving differential privacy, for various data types and scenarios. Until this work however, all previous work on differential privacy has been conducted on a ad-hoc basis, without a single, unifying codebase to implement results. In this work, we present the IBM Differential Privacy Library, a general purpose, open source library for investigating, experimenting and developing differential privacy applications in the Python programming language. The library includes a host of mechanisms, the building blocks of differential privacy, alongside a number of applications to machine learning and other data analytics tasks. Simplicity and accessibility has been prioritised in developing the library, making it suitable to a wide audience of users, from those using the library for their first investigations in data privacy, to the privacy experts looking to contribute their own models and mechanisms for others to use.

研究动机与目标

• 推动建立一个统一、可访问的差分隐私研究与应用的代码库的必要性。
• 介绍一个实现 DP 机制与模型的通用 Python 库。
• 展示与流行的 Python 数据科学工具（NumPy 和 scikit-learn）的集成，以便降低采用门槛。
• 通过示例和教程演示用法，以鼓励贡献和探索。

提出的方法

• 介绍 Diffprivlib 库及其设计目标（简单、易用、开源）。
• 描述三个主模块：mechanisms、models 和 tools，以及它们的互操作方式。
• 解释 API 设计（例如 set_epsilon、set_sensitivity、randomise）以及与 NumPy/Scikit-learn 约定的紧密对齐。
• 提供安装细节（pip install diffprivlib）以及与现有 Python 工具链的集成。
• 提供一个示例（Iris 上的 GaussianNB）以说明 DP 训练与评估。

实验结果

研究问题

• RQ1 diffprivlib 实现了哪些 DP 机制，它们如何组织？
• RQ2 用户如何使用该库训练和评估差分隐私机器学习模型？
• RQ3 diffprivlib 如何与 NumPy 和 scikit-learn 集成以促进 DP 实验？
• RQ4 该库提供了哪些指引来防止隐私泄露和确保正确配置？

主要发现

• Diffprivlib 提供了广泛的 DP 机制集合，分组在子模块中（例如 Laplace、Gaussian、Geometric）。
• 该库提供的 DP 启用的机器学习模型模仿 scikit-learn 的 API，几乎实现一行即可完成模型拟合。
• 包含用于差分隐私直方图和基本统计的工具，利用等同于 NumPy 的功能。
• 该软件包包括警告（PrivacyLeakWarning、DiffprivlibCompatibilityWarning），帮助用户避免隐私泄漏和不兼容性。
• 一个演示示例展示在 Iris 上的 DP Gaussian Naive Bayes，采用 80/20 训练/测试拆分，并在不同的 epsilon 值下进行 DP 精度分析。
• 该库以 MIT 许可发布，旨在广泛可访问，并欢迎社区贡献。