Assessing the impact of obfuscation on the process of identifying the author of a program code

Abstract: Various ways of obfuscating the source code can reduce the effectiveness of the applied models of identifying the author of the program code to random guessing. This article is devoted to assessing the influence of the fact of obfuscation of the source code on the process of identifying the author of a program using a hybrid neural network model. As part of the study, experiments were carried out with both interpreted and compiled programming languages. The results obtained indicate the stability of the previously proposed model of a hybrid neural network to obfuscation and the possibility of its application to solve the problem.

Keywords: author, source code, obfuscation, neural network, machine learning

For citation:
Kurtukova A. V., Svarovskaya E. E., Romanov A. S. Assessing the impact of obfuscation on the process of identifying the author of a program code. Doklady Tomskogo gosudarstvennogo universiteta sistem upravleniya i radioelektroniki, 2020, vol. 23, no. 2, pp. 50–54. DOI: 10.21293/1818-0442-2020-23-2-50-54