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 1709.04514cs.LGcs.CR
收藏

差分私有混合生成神经网络

Differentially Private Mixture of Generative Neural Networks

CrySyS Lab, BME-HIT(CrySyS实验室,布达佩斯技术与经济大学-网络系统与服务系);University College London(伦敦大学学院);INRIA(法国国家信息与自动化研究所)

Gergely Acs, Luca Melis, Claude Castelluccia, Emiliano De Cristofaro

摘要:

生成模型被广泛用于构建大量上下文丰富信息的应用程序中。然而,由于可能侵犯其数据被用于训练这些模型的个人的隐私,发布或共享生成模型并不总是可行的。在本文中,我们提出了一种新的技术,用于私下发布生成模型和由这些模型产生的整个高维数据集。我们用$k$生成神经网络的混合物对训练数据的生成器分布进行建模。它们一起训练,共同学习数据集的生成器分布。数据被分成$k$集群,使用一种新的差分私有内核$k$ -means,然后每个集群被赋予单独的生成神经网络,如限制玻尔兹曼机或变分自编码器,它们只在自己的集群上使用差分私有梯度下降进行训练。我们使用MNIST数据集以及呼叫详细记录和过境数据集来评估我们的方法,表明它产生了逼真的合成样本,这些样本也可用于准确计算任意数量的计数查询。

英文摘要:

Generative models are used in a wide range of applications building on large amounts of contextually rich information. Due to possible privacy violations of the individuals whose data is used to train these models, however, publishing or sharing generative models is not always viable. In this paper, we present a novel technique for privately releasing generative models and entire high-dimensional datasets produced by these models. We model the generator distribution of the training data with a mixture of $k$ generative neural networks. These are trained together and collectively learn the generator distribution of a dataset. Data is divided into $k$ clusters, using a novel differentially private kernel $k$-means, then each cluster is given to separate generative neural networks, such as Restricted Boltzmann Machines or Variational Autoencoders, which are trained only on their own cluster using differentially private gradient descent. We evaluate our approach using the MNIST dataset, as well as call detail records and transit datasets, showing that it produces realistic synthetic samples, which can also be used to accurately compute arbitrary number of counting queries.

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。

↑