Azerbaijan National Academy of Sciences

The article "Privacy-preserving deep learning algorithm for big personal data analysis" (DOI: https://doi.org/10.1016/j.jii.2019.07.002), co-authored by the Vice-president of ANAS, Director of ANASInstitute of Information Technology, academician Rasim Alguliyev, head of the institute’s department, corresponding member of ANAS, Doctor of Technical sciences Ramiz Aliguliyev and leading researcher, PhD in Technical sciences, associate professor Fargana Abdullaeva has been published in “Journal of Industrial Information Integration” has been published in “Elsevier” journal.

Privacy-preserving, the article proposes a method of deep learning for data analysis. The main goal of the method is to find sensitive data from a person’s personal data and transform it into non-sensitive data. To perform this process, it is proposed to use a two-block architecture. The architectural blocks are two modified neural networks “Sparse denoising Autoencoder” and “Convolutional Neural Network”, which performs the function of classification of transformed data. The Kulback-Leibler divergence function is included in the target function of the autoencoder neural network in order to minimize data transformation loss. The proposed method was tested on the data of Arrhythmia and Skoda, Cleveland medical dataset database "Heart Disease dataset".

As a result of a comparative analysis with existing methods, it was revealed that the proposed method is superior to them. This article was prepared as part of a grant of the project “Development of methods and algorithms for ensuring information security in a big data environment (“Big data”), some of their application” (Qrant No. EİF-KETPL-2-2015-1 (25) - 56/05/1) funded by the Presidential Science Development Fundation of the Republic of Azerbaijan.

Note that, the Journal of Industrial Information Integration journal is being indexed and referenced in such scientific databases as Web of Science, Emerging Sources Citation Index (ESCI), Ei Compendex, EBSCOhost, Scopus, INSPEC".

©All rights are re served. Citing to www.science.gov.az is necessary upon using news.