MODELING THE TRADEMARK GENERATION PROCESS USING NEURAL NETWORK

The paper discusses approaches to modeling the process of trademark creation using artificial neural networks. An analysis of modern generative architectures used in visual design, including diffusion models and generative adversarial networks (GANs), is presented. A conceptual model of an automated trademark design system is proposed, including the stages of training, generation, and originality assessment of images. The criteria for evaluating the aesthetic quality and uniqueness of generated signs are considered. The obtained results can be applied to optimize creative processes in branding and corporate identity development.