SHS Web of Conf.
Volume 44, 2018IV International Scientific Conference “The Convergence of Digital and Physical Worlds: Technological, Economic and Social Challenges” (CC-TESC2018)
|Number of page(s)||10|
|Published online||05 June 2018|
Application of deep neural networks for security analysis of digital infrastructure componentsa
Peter the Great St. Petersburg Polytechnic University, Institute of Computer Sciences and Technologies, Department of Computer Systems Information Security, 195251 Polytechnicheskaya st. 29, Russian Federation
* Corresponding author: email@example.com
In this article the authors give a consideration to a problem of detecting errors and vulnerabilities in software components of different digital devices. The article shows an ever-increasing criticality of this problem in the course of time related to development of modern concepts the Industrial Internet and the Industry 4.0. It gives an overview of modern approaches to application of methods of computer-assisted learning and artificial intellect in the sphere of cyber security, problems and prospects of application thereof. A new approach is offered to searching software vulnerabilities on the basis of application of deep learning. The approach is based on building semantically significant vector representations of software code and multistage instructing the deep neural network on revealing hierarchical abstractions in computer code testifying to presence of vulnerabilities. The authors describe specific features of the goal of analyzing software code for presence of vulnerabilities and proceeding thereof it is offered to use a neural network with long short-term memory (LSTM). In order to solve a problem of the learning set, the authors offer to use learning with transfer in case of building vector representations of instructions. The article also provides results of experimental investigations on application of offered solutions.
© The Authors, published by EDP Sciences, 2018
This is an Open Access article distributed under the terms of the Creative Commons Attribution License 4.0, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. (http://creativecommons.org/licenses/by/4.0/).
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.