arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 1807.01860cs.CRcs.LG
收藏

通过数据混淆保护隐私的机器学习

Privacy-preserving Machine Learning through Data Obfuscation

Princeton University(普林斯顿大学)

Tianwei Zhang, Zecheng He, Ruby B. Lee

摘要:

随着机器学习成为一种实践和商品,许多基于云的服务和框架被提供来帮助客户开发和部署机器学习应用。虽然将模型训练和服务任务外包到云中很普遍,但重要的是要保护训练数据集中敏感样本的隐私,并防止信息泄露给不受信任的第三方。过去的研究表明,恶意的机器学习服务提供商或最终用户可以很容易地从模型参数甚至只是模型输出中提取关于训练样本的关键信息。在本文中,我们提出了一种新的和通用的方法来保护隐私的训练数据在机器学习应用程序。具体来说,我们引入了一个模糊函数,并在将训练数据馈送到模型训练任务之前将其应用于训练数据。此函数向现有样本添加随机噪声,或使用新样本扩充数据集。通过这样做,隐藏了关于单个样本的属性或一组样本的统计属性的敏感信息。同时,从混淆数据集训练的模型仍然可以达到很高的准确率。通过这种方法,客户可以安全地向第三方提供商或最终用户披露数据或模型,而无需担心数据隐私。我们的实验表明,这种方法可以有效地击败现有的四种类型的机器学习隐私攻击在可以忽略不计的准确性成本。

英文摘要:

As machine learning becomes a practice and commodity, numerous cloud-based services and frameworks are provided to help customers develop and deploy machine learning applications. While it is prevalent to outsource model training and serving tasks in the cloud, it is important to protect the privacy of sensitive samples in the training dataset and prevent information leakage to untrusted third parties. Past work have shown that a malicious machine learning service provider or end user can easily extract critical information about the training samples, from the model parameters or even just model outputs. In this paper, we propose a novel and generic methodology to preserve the privacy of training data in machine learning applications. Specifically we introduce an obfuscate function and apply it to the training data before feeding them to the model training task. This function adds random noise to existing samples, or augments the dataset with new samples. By doing so sensitive information about the properties of individual samples, or statistical properties of a group of samples, is hidden. Meanwhile the model trained from the obfuscated dataset can still achieve high accuracy. With this approach, the customers can safely disclose the data or models to third-party providers or end users without the need to worry about data privacy. Our experiments show that this approach can effective defeat four existing types of machine learning privacy attacks at negligible accuracy cost.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑