{"type":"ScholarlyArticle","title":"Regression Model Fitting under Differential Privacy and Model Inversion Attack","authors":[{"name":"Yue Wang","slug":"yue-wang","given":"Yue","family":"Wang","affiliation":"University of North Carolina at Charlotte Charlotte , NC , USA"},{"name":"Cheng Si","slug":"cheng-si","given":"Cheng","family":"Si","affiliation":"University of Arkansas Fayetteville , AR , USA"},{"name":"Xintao Wu","slug":"xintao-wu","given":"Xintao","family":"Wu","affiliation":"University of Arkansas Fayetteville , AR , USA"}],"abstract":"Differential privacy preserving regression models guarantee protection against attempts to infer whether a subject was included in the training set used to derive a model. It is not designed to protect attribute privacy of a target individual when model inversion attacks are launched. In model inversion attacks, an adversary uses the released model to make predictions of sensitive attributes (used as input to the model) of a target individual when some background information about the target individual is available. Previous research showed that existing differential privacy mechanisms cannot effectively prevent model inversion attacks while retaining model efficacy. In this paper, we develop a novel approach which leverages the functional mechanism to perturb coefficients of the polynomial representation of the objective function but effectively balances the privacy budget for sensitive and non-sensitive attributes in learning the differential privacy preserving regression model. Theoretical analysis and empirical evaluations demonstrate our approach can effectively prevent model inversion attacks and retain model utility.","doi":null,"url":"https://ijcai.org/proceedings/2015/146","pdf":"https://ijcai.org/proceedings/2015/0146.pdf","markdown":"https://ijcai.org/proceedings/2015/146.md","bibtex":"https://ijcai.org/proceedings/2015/bibtex/146","volume":{"slug":"2015","title":"Proceedings of the Twenty-Fourth International Joint Conference on Artificial Intelligence","short_title":"IJCAI 2015","url":"https://ijcai.org/proceedings/2015/","year":2015,"location":"Buenos Aires, Argentina","dates":"25–31 July 2015","editors":"Qiang Yang and Michael Wooldridge","isbn":"978-1-57735-738-4","publisher":"International Joint Conferences on Artificial Intelligence","doi_prefix":""},"section":"Machine Learning","track":"Main Track","number":146,"pages":{"first":1003,"last":1008},"keywords":[],"videos":[],"references":[{"position":1,"raw":"[Barak et al., 2007] Boaz Barak, Kamalika Chaudhuri, Cyn- thia Dwork, Satyen Kale, Frank McSherry, and Kunal Tal- war. Privacy, accuracy, and consistency too: a holistic so- lution to contingency table release. In PODS, pages 273- 282. ACM, 2007.","title":"Privacy, accuracy, and consistency too: a holistic solution to contingency table release","authors":"Boaz Barak, Kamalika Chaudhuri, Cynthia Dwork, Satyen Kale, Frank Mcsherry, Kunal Talwar","year":"2007","venue":"PODS","doi":null,"url":null,"internal_url":null},{"position":2,"raw":"[Chaudhuri and Monteleoni, 2008] K. Chaudhuri and C. Monteleoni. Privacy-preserving logistic regression. In NIPS, pages 289-296.","title":"Privacy-preserving logistic regression","authors":"K Chaudhuri, C Monteleoni","year":null,"venue":"NIPS","doi":null,"url":null,"internal_url":null},{"position":3,"raw":"[Cormode, 2011] Graham Cormode. Personal privacy vs population privacy: learning to attack anonymization. In KDD, pages 1253-1261. ACM, 2011.","title":"Personal privacy vs population privacy: learning to attack anonymization","authors":"Graham Cormode","year":"2011","venue":"KDD","doi":null,"url":null,"internal_url":null},{"position":4,"raw":"[Dankar and El Emam, 2012] Fida Kamal Dankar and Khaled El Emam. The application of differential pri- vacy to health data. In Proceedings of the 2012 Joint EDBT/ICDT Workshops, pages 158-166. ACM, 2012.","title":"The application of differential privacy to health data","authors":"El Dankar, ; Emam, Kamal Fida, Khaled El Dankar, Emam","year":"2012","venue":"Proceedings of the 2012 Joint EDBT/ICDT Workshops","doi":null,"url":null,"internal_url":null},{"position":5,"raw":"[Ding et al., 2011] Bolin Ding, Marianne Winslett, Jiawei Han, and Zhenhui Li. Differentially private data cubes: optimizing noise sources and consistency. In SIGMOD, pages 217-228, 2011.","title":"Differentially private data cubes: optimizing noise sources and consistency","authors":"Bolin Ding, Marianne Winslett, Jiawei Han, Zhenhui Li","year":"2011","venue":"SIGMOD","doi":null,"url":null,"internal_url":null},{"position":6,"raw":"[Dwork et al., 2006] Cynthia Dwork, Frank McSherry, Kobbi Nissim, and Adam Smith. Calibrating noise to sen- sitivity in private data analysis. Theory of Cryptography, pages 265-284, 2006.","title":"Calibrating noise to sensitivity in private data analysis","authors":"Cynthia Dwork, Frank Mcsherry, Kobbi Nissim, Adam Smith","year":"2006","venue":"Theory of Cryptography","doi":null,"url":null,"internal_url":null},{"position":7,"raw":"[Fredrikson et al., 2014] Matthew Fredrikson, Eric Lantz, Somesh Jha, Simon Lin, David Page, and Thomas Risten- part. Privacy in pharmacogenetics: An end-to-end case study of personalized warfarin dosing. In Proceedings of the 23rd USENIX Conference on Security Symposium, pages 17-32, 2014.","title":"Privacy in pharmacogenetics: An end-to-end case study of personalized warfarin dosing","authors":"Matthew Fredrikson, Eric Lantz, Somesh Jha, Simon Lin, David Page, Thomas Ristenpart","year":"2014","venue":"Proceedings of the 23rd USENIX Conference on Security Symposium","doi":null,"url":null,"internal_url":null},{"position":8,"raw":"[Hay et al., 2009] Michael Hay, Chao Li, Gerome Miklau, and David Jensen. Accurate estimation of the degree dis- tribution of private networks. In ICDM, pages 169-178. IEEE, 2009.","title":"Accurate estimation of the degree distribution of private networks","authors":"Hay","year":"2009","venue":"ICDM","doi":null,"url":null,"internal_url":null},{"position":9,"raw":"[Hay et al., 2010] Michael Hay, Vibhor Rastogi, Gerome Miklau, and Dan Suciu. Boosting the Accuracy of Dif- ferentially Private Histograms Through Consistency. Pro- ceedings of the VLDB Endowment, 3(1), 2010.","title":"Boosting the Accuracy of Differentially Private Histograms Through Consistency","authors":"Michael Hay, Vibhor Rastogi, Gerome Miklau, Dan Suciu","year":"2010","venue":"Proceedings of the VLDB Endowment","doi":null,"url":null,"internal_url":null},{"position":10,"raw":"[Kifer and Machanavajjhala, 2011] Daniel Kifer and Ash- win Machanavajjhala. No free lunch in data privacy. In SIGMOD, pages 193-204, 2011.","title":"Daniel Kifer and Ashwin Machanavajjhala. No free lunch in data privacy","authors":"Machanavajjhala Kifer","year":"2011","venue":"SIGMOD","doi":null,"url":null,"internal_url":null},{"position":11,"raw":"[Lee and Clifton, 2012] Jaewoo Lee and Chris Clifton. Dif- ferential identifiability. In KDD, pages 1041-1049, 2012.","title":"Differential identifiability","authors":"Jaewoo Lee, Chris Clifton","year":"2012","venue":"KDD","doi":null,"url":null,"internal_url":null},{"position":12,"raw":"[Lichman, 2013] M. Lichman. UCI machine learning repos- itory, 2013.","title":"UCI machine learning repository","authors":"M Lichman","year":"2013","venue":null,"doi":null,"url":null,"internal_url":null},{"position":13,"raw":"[McSherry and Mironov, 2009] Frank McSherry and Ilya Mironov. Differentially Private Recommender Systems. In KDD. ACM, 2009.","title":"Differentially Private Recommender Systems","authors":"Mironov Mcsherry, Frank Mcsherry, Ilya Mironov","year":"2009","venue":"KDD. ACM","doi":null,"url":null,"internal_url":null},{"position":14,"raw":"[McSherry, 2009] Frank McSherry. Privacy integrated queries: an extensible platform for privacy-preserving data analysis. In SIGMOD, pages 19-30. ACM, 2009.","title":"Privacy integrated queries: an extensible platform for privacy-preserving data analysis","authors":"Frank Mcsherry, Mcsherry","year":"2009","venue":"SIGMOD","doi":null,"url":null,"internal_url":null},{"position":15,"raw":"[Mir and Wright, 2009] Darakhshan Mir and Rebecca Wright. A differentially private graph estimator. In ICDMW, pages 122-129. IEEE, 2009.","title":"Darakhshan Mir and Rebecca Wright. A differentially private graph estimator","authors":"Wright Mir","year":"2009","venue":"ICDMW","doi":null,"url":null,"internal_url":null},{"position":16,"raw":"[Rastogi et al., 2009] Vibhor Rastogi, Miachel Hay, Gerome Miklau, and Dan Suciu. Relationship privacy: Output per- turbation for queries with joins. In PODS, pages 107-116. ACM, 2009.","title":"Relationship privacy: Output perturbation for queries with joins","authors":"Vibhor Rastogi, Miachel Hay, Gerome Miklau, Dan Suciu","year":"2009","venue":"PODS","doi":null,"url":null,"internal_url":null},{"position":17,"raw":"[Sala et al., 2011] Alessandra Sala, Xiaohan Zhao, Christo Wilson, Haitao Zheng, and Ben Zhao. Sharing graphs using differentially private graph models. In ACM SIG- COMM, pages 81-98. ACM, 2011.","title":"Sharing graphs using differentially private graph models","authors":"Alessandra Sala, Xiaohan Zhao, Christo Wilson, Haitao Zheng, Ben Zhao","year":"2011","venue":"ACM SIG-COMM","doi":null,"url":null,"internal_url":null},{"position":18,"raw":"[Wang and Wu, 2013] Yue Wang and Xintao Wu. Preserving differential privacy in degree-correlation based graph gen- eration. Transactions on Data Privacy, 6:127-145, 2013.","title":"Preserving differential privacy in degree-correlation based graph generation","authors":"Yue Wang, Xintao Wu","year":"2013","venue":"Transactions on Data Privacy","doi":null,"url":null,"internal_url":null},{"position":19,"raw":"[Wang et al., 2012] Yue Wang, Xintao Wu, Jun Zhu, and Yang Xiang. On learning cluster coefficient of private net- works. In ASONAM, 2012.","title":"On learning cluster coefficient of private networks","authors":"Yue Wang, Xintao Wu, Jun Zhu, Yang Xiang","year":"2012","venue":"ASONAM","doi":null,"url":null,"internal_url":null},{"position":20,"raw":"[Wang et al., 2013] Yue Wang, Xintao Wu, and Leting Wu. Differential privacy preserving spectral graph analysis. In PAKDD (2), pages 329-340, 2013.","title":"Differential privacy preserving spectral graph analysis","authors":"Yue Wang, Xintao Wu, Leting Wu","year":"2013","venue":"PAKDD (2)","doi":null,"url":null,"internal_url":null},{"position":21,"raw":"[Xiao et al., 2010] Xiaokui Xiao, Guozhang Wang, and Jo- hannes Gehrke. Differential privacy via wavelet trans- forms. In ICDE, pages 225-236. IEEE, 2010.","title":"Differential privacy via wavelet transforms","authors":"Xiao","year":"2010","venue":"ICDE","doi":null,"url":null,"internal_url":null},{"position":22,"raw":"[Ying et al., 2013] Xiaowei Ying, Xintao Wu, and Yue Wang. On linear refinement of differential privacy- preserving query answering. In PAKDD (2), pages 353- 364, 2013.","title":"On linear refinement of differential privacypreserving query answering","authors":"Xiaowei Ying, Xintao Wu, Yue Wang","year":"2013","venue":"PAKDD","doi":null,"url":null,"internal_url":null},{"position":23,"raw":"[Zhang et al., 2012] Jun Zhang, Zhenjie Zhang, Xiaokui Xiao, Yin Yang, and Marianne Winslett. Functional mechanism: regression analysis under differential privacy. Proceedings of the VLDB Endowment, 5(11):1364-1375, 2012.","title":"Functional mechanism: regression analysis under differential privacy","authors":"Zhang","year":"2012","venue":"Proceedings of the VLDB Endowment","doi":null,"url":null,"internal_url":null}],"license_note":"Copyright International Joint Conferences on Artificial Intelligence."}