Abstract:

A novel approach to supervised face generation using open-source generative models including StyleGAN2 and Ridge Regression is presented. A methodology that extends StyleGAN2 to control facial characteristics such as age, race, gender, facial expression, and hair attributes is developed, and an extensive dataset of human faces with attribute annotations is utilized. The faces were encoded in 256-dimensional latent space using the StyleGAN2 encoder, resulting in a set of characteristic latent codes. We applied the t-SNE algorithm to cluster these feature-based codes, demonstrated the ability to control face generation, and subsequently trained Ridge regression models for each dimension of the latent codes using the labeled features. When decoded using StyleGAN2, the resulting codes successfully reconstructed face images while maintaining the association with the input features. The developed approach provides an easy and efficient way to supervised face generation using existing generative models such as StyleGAN2, and opens up new possibilities for different application areas.

Abstract:

The article describes a typical object-oriented WEB-system designed for storing and providing various reference and statistical data on the scientific works of employees of an institution (group of institutions), developed by specialists of the JSCC RAS. The system contains information about publications of employees and reports made by them at scientific conferences, symposiums, and seminars. The system is focused on working with objects belonged to classes connected between each other, such as "author", "organization", "publication", "report", "event". The metadata profile of objects of each class includes attributes that are necessary to get detailed information about both an individual object of this class and a group of objects associated with the specified attribute values of objects of other classes. For example, you have to get a list of articles by employees of a given organization published articles in a given journal for a given period of time. A distinctive feature of the system is the introduced concept of "equivalent" objects. Such objects are "persons" corresponding to the same author with different spellings of the last name in the bibliographic descriptions of publications; organizations with different versions of names; articles which are published without changes in different languages. This article describes in detail the features of the system, its user interface, and provides examples of performing specific queries.

Abstract: Methods of the semantic documents parsing in digital control system of scientific collections, including electronic journals, offered. The methods of processing documents containing mathematical formulas and methods for the conversion of documents from the OpenXML-format in ТеХ-format considered. The search algorithm for the mathematical formulas in the collections of documents stored in OpenXML-format designed. The algorithm is implemented as online-service on platform science.tatarstan.

Abstract:

The Common Digital Space of Scientific Knowledge (CDSSK) is a digital information environment aggregating heterogeneous information related to various aspects of scientific knowledge. One of the important functions of the CDSSK is to provide information for solving artificial intelligence problems, which makes it necessary to support data in a structure that complies with the rules of the semantic WEB. The features of the CDSSK are, on the one hand, the polythematics and heterogeneity of content elements, on the other hand, the high dynamics of the emergence of new types of objects and connections between them, which is due to the specifics of the development of science. At the same time, it should be possible to navigate through heterogeneous space resources using semantic links between them. The possibilities of the CDSSK are largely determined by the structure of the ontology of space, the model of which is proposed in this paper. Within the framework of the model, the hierarchical structuring of the CDSSK ontology is carried out; such elements as "subspace", "class of objects", "object", "attributes of an object", three types of pairwise relations of objects or attributes (universal, quasi-universal and specific) are distinguished and defined. The structure of each elements type is determined by a "reference book" of a unified type; specific values of attributes and relationships are contained in dictionaries of a unified structure. A class of "Formats" objects describing the rules for the formation of attributes and values of relationships is allocated. The formalization of CDSSK reference books and dictionaries representations is proposed. The proposed model allows you to simply add new types of objects, of their pairwise relationships and attributes to the space, as needed.