Персона: Кулик, Сергей Дмитриевич
Email Address
Birth Date
Результаты поиска
Effective scientific personnel training in the field of modern computer technologies for the implementation of advanced research projects of the Megascience class
© Published under licence by IOP Publishing Ltd.Successful research projects of the Megascience class usually require a well-trained team of scientists from various fields of knowledge. These scientists must be high-skilled experts. Each member of a team like that must have the necessary, specialized cross-industry skills, for example, in such areas as artificial intelligence, convolutional neural networks, specialized intelligent search engines, and full-text analysis. One of the key aspects of effective personnel training for successful implementation of Megascience projects into reality is the acquisition by students professional skills, abilities, and knowledge to use tools of modern scientific technologies, containing, for example, libraries of programs (functions). In particular, convolutional neural networks and intelligent search systems can be applied in various research projects in the field of physics, chemistry, biology, and medicine, for example, in telemedicine, for effective decision-making in diagnosing a patient. Therefore, understanding the principles of neural networks and intelligent search systems is a necessary competence of researchers working in the framework of Megascience projects. Classic search engines are based on indexing the textual information of the database that is being searched. Intelligent search engines can improve the search experience through intelligent data processing, including using convolutional neural networks. This report examines practical examples and areas of the successful application of convolutional neural networks and information systems in practice.
Scientific personnel training in convolutional neural networks for the implementation of research projects of the MegaScience class
© Published under licence by IOP Publishing Ltd.Megascience projects require an all-inclusive interdisciplinary approach. Because of that scientific personnel engaged in projects of this type must possess relevant required interdisciplinary skills. Artificial intelligence in particular convolutional neural networks has a wide range of applications, and it could be used to solve complicated problems in all kinds of various fields of science. Thus, the understanding of the principles of neural networks' working is a necessary skill for scientific personnel. In this paper, we're considering practical examples and fields of applications of neural networks in the real world.
Experiments with Neural Net Object Detection System YOLO on Small Training Datasets for Intelligent Robotics
© 2020, Springer Nature Switzerland AG.In this paper we’ve conducted multiple experiments with modern object detection system YOLO. Object detection systems are fundamental to many robotics tasks. Recognition algorithms involving object detection are often part of various intelligence systems for robots. Training object detection systems usually requires waste amounts of training data which can be expensive and time-consuming. In this paper we’ve conducted several experiments with YOLO on small training datasets investigating YOLO’s capacity to train on small number of examples. We measured accuracy metrics for object detector depending on the size of training dataset, compared training process of full and smaller versions of YOLO and their speed. Gathered information will be used for creating visual factographic intelligence system for robots. YOLO (You Only Look Once) is a special intelligent technology for computer vision techniques. Our results are useful for industry professionals and students from a broad range of disciplines related to robotics, intelligent technologies and other fields.
Recognition Algorithm for Biological and Criminalistics Objects
© 2020, Springer Nature Switzerland AG.This paper describes the results of a work to develop an algorithm for analyzing images of embossed impressions in paper documents under oblique lighting. The described algorithm could also be used for recognition of similarly-structured objects, for example, some of biological structures. This type of analysis is necessary during forensic analysis of certain security features of paper documents. Part of this analysis is determining to which category new, uncategorized impression belongs to. This research explores the potential for automation of this task using neural networks. The core element of the algorithm is a neural network which determines the similarity between two embossed impressions. The paper describes the structure of the algorithm, a method for creating an image database for training and testing, as well as testing results for proposed algorithm.