基于迁移学习的敏感数据隐私保护方法

被引:8
作者
付玉香 [1 ]
秦永彬 [1 ,2 ]
申国伟 [1 ,2 ]
机构
[1] 贵州大学计算机科学与技术学院
[2] 贵州大学贵州省公共大数据重点实验室
基金
国家自然科学基金重大研究计划;
关键词
差分隐私; 迁移学习; 模型攻击; 敏感数据; 隐私保护;
D O I
10.16337/j.1004-9037.2019.03.006
中图分类号
TP309 [安全保密]; TP181 [自动推理、机器学习];
学科分类号
081201 ; 0839 ; 1402 ;
摘要
机器学习涉及一些隐含的敏感数据,当受到模型查询或模型检验等模型攻击时,可能会泄露用户隐私信息。针对上述问题,本文提出一种敏感数据隐私保护"师徒"模型PATE?T,为机器学习模型的训练数据提供强健的隐私保证。该方法以"黑盒"方式组合了由不相交敏感数据集训练得到的多个"师父"模型,这些模型直接依赖于敏感训练数据。"徒弟"由"师父"集合迁移学习得到,不能直接访问"师父"或基础参数,"徒弟"所在数据域与敏感训练数据域不同但相关。在差分隐私方面,攻击者可以查询"徒弟",也可以检查其内部工作,但无法获取训练数据的隐私信息。实验表明,在数据集MNIST和SVHN上,本文提出的隐私保护模型达到了隐私/实用准确性的权衡,性能优越。
引用
收藏
页码:422 / 431
页数:10
相关论文
共 30 条
[11]  
Gradient-based learning applied to document recognition. LeCun L,Bottou L,Bengio Y,Haffner P. Proceedings of Tricomm . 1998
[12]  
K-anonymity: a model for protecting privacy. Sweeney L. International Journal on Uncertainty,Fuzziness and Knowledge-Based Systems . 2002
[13]  
Contour Detection and Hierarchical Image Segmentation. Pablo Arbelaez,Michael Maire,Charless Fowlkes,Jitendra Malik. IEEE Transactions on Pattern Analysis and Machine Intelligence . 2011
[14]  
Domain adversarial training of neural networks. Ganin Y,Ustinova E,Ajakan H,Germain P,Larochell H,Laviolette F,Marchand M,Lempitsky V. Journal of Machin Learning Research . 2016
[15]  
Deep learning requires rethinking generalization. Zhang C,Bengio S,Hardt M, et al. International Conference on Learning Representation (ICLP) . 2017
[16]  
Multiparty differential privacy via aggregation of locally trained classifiers. Pathak M,Rane S,Raj B. Advances in Neural Information Processing Systems . 2010
[17]  
Learning privately from multiparty data. Hamm J,Cao Y,Belkin M. International Conference on Machine Learning . 2016
[18]  
A semi?supervised learning approach to differential privacy. Jagannathan G,Monteleoni C,Pillaipakkamnatt K. IEEE 13th International Conference on Data Mining Workshops (ICDMW) . 2013
[19]  
Semi?supervised knowledge transfer for deep learning from private training data. Papernot N,Abadi M,Erlingsson U, et al. International Conference on Learning Representations . 2016
[20]  
Deep transfer learning with joint adaptation networks. Long M,Zhu H,Wang J, et al. International Conference on Machine Learning . 2017