Year: 2023
Type: conference-paper
Abstract: The article considers the challenges of cybersecurity that appear at the current stage of information technology development, along with the introduction of artificial intelligence and machine learning. Based on the generated information about the production processes and ways of filling in the knowledge base and the database, a model for creating a knowledge base and expert simulation developed, which made it possible to make an optimal assessment of expert knowledge, and in the event of their contradiction, use the knowledge of the simulated expert. Protection against attacks by competitors provided with the replacement of correct information. Modeling of the system's operation under the conditions of a competitive FGM attack carried out. Analysis of the transient characteristics showed that as the epsilon index increases, distortion of the samples occurs, leading to the loss of data important for the operation of the system. In turn, the number of copies for which a decision made decreases, and thus the accuracy of recognizing error signals and disturbances obtained during the operation of printing devices partially reduced. Considering that, the samples that have not passed the inspection not admitted to the decision-making stage on them and do not affect the accuracy. The system works in normal mode and minimizes or even eliminates the influence of competition, depending on the epsilon, designed the printing house management system, thanks to a simulated expert and protection against competition attacks, will ensure the continuity of the production plant processes. Experiments carried out that show the effectiveness of the development in both increasing the accuracy of the classifier and ensuring its reliable operation in the conditions of a competitive attack.
Authors: Bohdan Durnyak, Petro Shepita, Lyubov Tupychak
Institution: Ukrainian Academy of Printing
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: bronze
OpenAlex ID: W7166311715