Neural network obfuscation for computations over encrypted data | Applied Discrete Mathematics. Supplement. 2020. № 13. DOI: 10.17223/2226308X/13/25

Neural network obfuscation for computations over encrypted data

An approach to neural network cryptographic obfusca-tion of computations is proposed. Applying the previously obtained results on the property of strict obfuscation of indistinguishability for a neural network approximator, we propose to use neural networks to perform arithmetic and other operations on encrypted data, thus realizing the idea of using homomorphic encryption to perform trusted computations in an untrusted environment. The cryptographic properties of this mechanism are evaluated and compared with traditional approaches to encryption based on the secret key. The advantages and disadvantages of neural networks in relation to the problem of obfuscation and processing of encrypted data are discussed.

Download file

Counter downloads: 83

Keywords

искусственная нейронная сеть, обфускация, гомоморфное шифрование, оценка стойкости, artificial neural network, obfuscation, homomorphic encryption, secrecy estimation

Authors

Name	Organization	E-mail
Eliseev V. L.	OJSC Infotecs; National Research University "Moscow Power Engineering Institute"	vlad-eliseev@mail.ru

Всего: 1

References

Николенко С., Кадурин А., Архангельская Е. Глубокое обучение. Погружение в мир нейронных сетей. СПб.: Питер, 2020.

Алексеев Д. В. Приближение функций нескольких переменных нейронными сетями // Фундаментальная и прикладная математика. 2009. Т. 15. №3. С. 9-21.

Хайкин С. Нейронные сети: полный курс. 2-е изд. М.: Вильямс, 2008.

Focardi R. and Luccio F. L. Neural cryptanalysis of classical ciphers // Proc. ICTCS. 2018. http://ceur-ws.org/Vol-2243/paper10.pdf

Danziger M. and Henriques M. A. A. Improved cryptanalysis combining differential and artificial neural network schemes // Intern. Telecommun. Symp. (ITS). Sao Paulo, 2014. P. 1-5.

Turcanik M. Using recurrent neural network for hash function generation // Intern. Conf. Appl. Electronics (AE). Pilsen, 2017. P. 1-4.

Lian S., Sun J., and Wang Z. One-way hash function based on neural network // arXiv:0707.4032. 2007.

Karras D. A. and Zorkadis V. On neural network techniques in the secure management of communication systems through improving and quality assessing pseudorandom stream generators // Neural Networks. 2003. V. 16. Iss. 5-6. P. 899-905.

Kanter I., Kinzel W., and Kanter E. Secure exchange of information by synchronization of neural networks // Europhys. Lett. 2002. No. 57. P. 141-147.

Klimov A., Mityagin A., and Shamir A. Analysis of neural cryptography // LNCS. 2002. V. 2502. P. 288-298.

Coutinho M., De Oliveira A. R., Borges F., et al. Learning perfectly secure cryptography to protect communications with adversarial neural cryptography // Sensors. 2018. No. 18. Article 1306. https://pubmed.ncbi.nlm.nih.gov/29695066/

Abadi M. and Andersen D. G. Learning to protect communications with adversarial neural cryptography // arXiv:1610.06918. 2016.

Xie P., Bilenko M., Finley T., et al. Crypto-nets: Neural networks over encrypted data // arXiv:1412.6181. 2014.

Hecht-Nielsen R. Kolmogorov's mapping neural network existence theorem // IEEE First Annual Int. Conf. on Neural Networks, San Diego, 1987. V. 3. P. 11-13.

Dathathri R., Saarikivi O., Chen H., et al. CHET: an optimizing compiler for fully-homomorphic neural-network inferencing // Proc. PLDI 2019. N.Y.: ACM, 2019. P. 142-156.

Елисеев В. Л. Искусственные нейронные сети как механизм обфускации вычислений // Прикладная дискретная математика. Приложение. 2019. № 12. С. 165-169.

Фергюсон Н., Шнайер Б. Практическая криптография. М.: Диалектика, 2005.