Deepfake Detection Models Based on Machine Learning Technologies

Abstract

The paper is devoted to efficiency evaluation of modern deepfake detection models based on convolutional neural networks (CNN). In the context of rapid development of digital technologies and increasing volume of information on the internet, the relevance of detecting fake images, videos, and textual materials becomes increasingly significant. Fake content, spread through social networks and other platforms, can have serious consequences, ranging from individual malicious attacks to manipulations of public opinion on a global level. We have built and trained several models for detecting fake content using convolutional neural networks. The training was performed using Deepfake Detection Challenge Dataset. During the study, we carried out the comparative analysis of the created models. Obtained results were compared with a number of recent publications.