Abstract
We study the maximal mutual information about a random variable Y (representing non-private information) displayed through an additive Gaussian channel when guaranteeing that only \(\varepsilon \) bits of information is leaked about a random variable X (representing private information) that is correlated with Y. Denoting this quantity by \(g_\varepsilon (X,Y)\), we show that for perfect privacy, i.e., \(\varepsilon =0\), one has \(g_0(X,Y)=0\) for any pair of absolutely continuous random variables (X, Y) and then derive a second-order approximation for \(g_\varepsilon (X,Y)\) for small \(\varepsilon \). This approximation is shown to be related to the strong data processing inequality for mutual information under suitable conditions on the joint distribution \(P_{XY}\). Next, motivated by an operational interpretation of data privacy, we formulate the privacy-utility tradeoff in the same setup using estimation-theoretic quantities and obtain explicit bounds for this tradeoff when \(\varepsilon \) is sufficiently small using the approximation formula derived for \(g_\varepsilon (X,Y)\).
S. Asoodeh—This work was supported in part by NSERC of Canada.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
We will see in the next section that this holds in the estimation-theoretic formulation of privacy, i.e., the Gaussian case is the worst case when the privacy filter is an additive Gaussian channel and the utility and privacy are measured as \({\mathsf {mmse}}(Y|Z_\gamma )\) and \({\mathsf {mmse}}(X|Z_\gamma )\), respectively.
References
Asoodeh, S., Alajaji, F., Linder, T.: Notes on information-theoretic privacy. In: Proceedings of the 52nd Annual Allerton Conference on Communication, Control, and Computing, pp. 1272–1278, September 2014
Asoodeh, S., Diaz, M., Alajaji, F., Linder, T.: Information extraction under privacy constraints. Information 7(1) (2016). http://www.mdpi.com/2078-2489/7/1/15
Calmon, F.P., Varia, M., Médard, M., Christiansen, M.M., Duffy, K.R., Tessaro, S.: Bounds on inference. In: Proceedings of the 51st Annual Allerton Conference on Communication, Control, and Computing, pp. 567–574, October 2013
Asoodeh, S., Alajaji, F., Linder, T.: Privacy-aware MMSE estimation. In: Proceedings of the IEEE International Symposium on Information (ISIT), July 2016. arXiv:1511.02381v3
Makhdoumi, A., Fawaz, N.: Privacy-utility tradeoff under statistical uncertainty. In: Proceedings of the 51st Allerton Conference on Communication, Control, and Computing, pp. 1627–1634, October 2013
Calmon, F.P., Makhdoumi, A., Médard, M.: Fundamental limits of perfect privacy. In: Proceedings of the IEEE International Symposium on Information Theory (ISIT), pp. 1796–1800 (2015)
Berger, T., Yeung, R.: Multiterminal source encoding with encoder breakdown. IEEE Trans. Inf. Theor. 35(2), 237–244 (1989)
Reed, I.S.: Information theory and privacy in data banks. In: Proceedings of the National Computer Conference and Exposition, ser. AFIPS 1973, pp. 581–587. ACM, New York (1973)
Yamamoto, H.: A source coding problem for sources with additional outputs to keep secret from the receiver or wiretappers. IEEE Trans. Inf. Theor. 29(6), 918–923 (1983)
Sankar, L., Rajagopalan, S., Poor, H.: Utility-privacy tradeoffs in databases: an information-theoretic approach. IEEE Trans. Inf. Forensics Secur. 8(6), 838–852 (2013)
Asoodeh, S., Alajaji, F., Linder, T.: On maximal correlation, mutual information and data privacy. In: Proceedings of the IEEE 14th Canadian Workshop on Information Theory (CWIT), pp. 27–31, June 2015
Rebollo-Monedero, D., Forne, J., Domingo-Ferrer, J.: From t-closeness-like privacy to postrandomization via information theory. IEEE Trans. Knowl. Data Eng. 22(11), 1623–1636 (2010)
Calmon, F.P.: Information-theoretic metrics for security and privacy. Ph.D. dissertation, MIT, September 2015
Goldwasser, S., Micali, S.: Probabilistic encryption. J. Comput. Syst. Sci. 28(2), 270–299 (1984)
Anantharam, V., Gohari, A., Kamath, S., Nair, C.: On maximal correlation, hypercontractivity, the data processing inequality studied by Erkip, Cover. Preprint, arXiv:1304.6133v1 (2014)
Guo, D., Shamai, S., Verdú, S.: Mutual information and minimum mean-square error in Gaussian channels. IEEE Trans. Inf. Theor. 51(4), 1261–1282 (2005)
Guo, D., Wu, Y., Shamai, S., Verdú, S.: Estimation in Gaussian noise: properties of the minimum mean-square error. IEEE Trans. Inf. Theor. 57(4), 2371–2385 (2011)
Rényi, A.: On measures of dependence. Acta Mathe. Acad. Scient. Hung. 10(3), 441–451 (1959)
Gebelein, H.: Das statistische problem der korrelation als variations- und eigenwert-problem und sein zusammenhang mit der ausgleichungsrechnung. Zeitschrift f ur angew. Math. und Mech. 21, 364–379 (1941)
Sarmanov, O.: The maximum correlation coefficient (nonsymmetric case). Dokl. Akad. Nauk SSSR 120(4), 715–718 (1958)
Cover, T.M., Thomas, J.A.: Elements of Information Theory. Wiley-Interscience, Hoboken (2006)
Polyanskiy, Y., Wu, Y.: Dissipation of information in channels with input constraints. IEEE Trans. Inf. Theor. 62(1), 35–55 (2016)
Wu, Y., Verdú, S.: Functional properties of minimum mean-square error and mutual information. IEEE Trans. Inf. Theor. 58(3), 1289–1301 (2012)
Prelov, V.V.: Capacity of communication channels with almost Gaussian noise. Teor. Veroyatnost. i Primenen. 33(3), 433–452 (1988)
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Appendices
A Connection Between Mutual Information and Non-Gaussianness
For any pair of random variables (U, V) with \(I(U; V){<\infty }\), let \(P_{V|U}(\cdot |u)\) be the conditional density of V given \(U=u\). Then, we have
where \((U_{\mathsf {G}}, V_{\mathsf {G}})\) is a pair of Gaussian random variable having the same means, variances and correlation coefficient as (U, V), and \(P_{V_{\mathsf {G}}|U_{\mathsf {G}}}(\cdot |u)\) is the conditional density of \(V_{\mathsf {G}}\) given \(U_{\mathsf {G}}=u\), and the quantity D(V|U) is defined in (34). Replacing U and V with X and \(Z_\gamma \), respectively, the decomposition (37) allows us to conclude that
and therefore, if \(Y=Y_{\mathsf {G}}\) is Gaussian, we have
from which we conclude that when Y is Gaussian then \(I(X; Z_\gamma )\le \varepsilon \) implies that \(I(X_{\mathsf {G}}; Z_\gamma )\le \varepsilon \) and hence \(g_\varepsilon (X, Y_{\mathsf {G}})\le g_\varepsilon (X_{\mathsf {G}}, Y_{\mathsf {G}})\).
B Completion of the Proof of Theorem 4
Lemma 1
For Gaussian \(X_{\mathsf {G}}\) and absolutely continuous Y with unit variance, we have
Proof
Let E be an auxiliary random variable defined as
for some real number \(M>0\). Note that
where (a) follows from the fact that \(h(Z_\gamma |X_{\mathsf {G}}=x, E=0)\ge h(N_{\mathsf {G}})\).
Prelov [24] showed that for any random variable Y such that
for some \(\alpha >0\), then
where \(o(\gamma )\) term depends only on K. Since \(Y|\{E=1\}\) satisfies (39), we can use (40) to evaluate \(h(Z_\gamma |X_{\mathsf {G}}=x, E=1)\) in (38) which yields
Note that since \({\mathsf {var}}(Y){<\infty }\) and \(X_{\mathsf {G}}\) has a positive density, \({\mathsf {var}}(Y|X_{\mathsf {G}}=x){<\infty }\) for almost all x (except for x in a set of zero Lebesgue measure). Hence, we can choose L sufficiently large such that for any given \(\delta >0\),
and
Therefore, invoking the inequality \(\log (1+u)\le u\) for \(u>0\), we can write
from which and the fact the \(\delta \) is arbitrarily small the result follows. \(\square \)
Rights and permissions
Copyright information
© 2016 Springer International Publishing AG
About this paper
Cite this paper
Asoodeh, S., Alajaji, F., Linder, T. (2016). Almost Perfect Privacy for Additive Gaussian Privacy Filters. In: Nascimento, A., Barreto, P. (eds) Information Theoretic Security. ICITS 2016. Lecture Notes in Computer Science(), vol 10015. Springer, Cham. https://doi.org/10.1007/978-3-319-49175-2_13
Download citation
DOI: https://doi.org/10.1007/978-3-319-49175-2_13
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-319-49174-5
Online ISBN: 978-3-319-49175-2
eBook Packages: Computer ScienceComputer Science (R0)