STUDYING HIGHER MATHEMATICS MODELS IN ANALYZING MULTIMODAL INFORMATION

Abstract

Summarizing the diversity of different models of higher mathematics in problems of representing multimodal information in a vector space, the authors note classification models, recognition models, multimodal hashing models and discriminative correlation analysis models. The article solves problems of determining the basic principles of displaying multimodal information in higher mathe- matics models, considers the most popular methods of higher mathematics in solving problems of classification models of multimodal information, and analyzes the most effective models of higher mathematics in displaying multimodal information. The scientific novelty of the article lies in the idea of representing multimodal information in higher mathematics models by the SVM method as a classifier in the search for an equation of a separating hyperplane in the feature space that optimally separates recognition classes in space. The study shows that multimodal information can be represented in higher mathematics models within a vector space using various methods. It is shown that it is effective to use multimodal hashing and discriminative correlation analysis to minimize semantic similarity between different modalities and extract discriminative representations from multiple data sources. It is also found that combining multimodal data in the form of textual and visual representations of information within a summary is effective in representing multimodal data.