| 2012-2016 | Development of intelligent information technologies for perception, recognition and interpretation of the environmental objects on the basis of formal models of imaginative thinking |
| Funded by National Academy of Sciences of Ukraine | |
| Modern information technologies require complex intelligent data processing in different domains. This project will develop and implement intelligent information technologies for perception, recognition and interpretation of the two most important and widely used types of information: text data in the form poorly structured text information and numerical data as signal patterns obtained by indirect measurements. Will be developed and implemented by software tools: - Methods for the perception and recognition of textual information structure by its visual patterns; - Methods for extracting and interpreting semantically meaningful textual information; - Methods for effective recovery of signals by their patterns obtained by remote sensing; - Methods of representation and conversion of signal patterns to improve the efficiency of processing. Results of the development are expected to be applied in the following areas: - Internet resources, search engines, information portals, data warehousing, data analysis systems, intelligent information management, powerful processing data streams from the Internet. - Geophysical exploration (electrical, magnetic, seismic, gravity), spectrometry (in different regions of the spectrum, such as gamma-ray spectrometry), detecting the presence of certain chemical elements in objects, medical diagnostics (imaging, tomography), nondestructive testing, remote monitoring. | |
| 2011-2015 | Development and research of methods for neural information representation and processing in intelligent information technologies |
| Funded by National Academy of Sciences of Ukraine | |
| The purpose of research on the topic is to improve a wide range of tasks on the basis of learning from examples and reasoning through the development of methods and algorithms for neural randomized distributed representation and processing of information. As a result of the research will be established conceptual framework, methods and algorithms for data representation and processing that scale well and will be useful for intelligent information technology, learning and applying the reasoning by example and to process large amounts of information. Will be developed and the methods of randomized vector representation of different kinds of information - the results of measurements of word, descriptions of situations, etc. Will be developed methods of information processing, which has the structure of the sequences, as well as information from a complex hierarchical structure. Will be developed methods for solving measurement information processing (such as a discrete ill-posed inverse problem) using randomized transformation, and study their computational efficiency and accuracy of the solutions. Proposed methods of teaching through the use of examples and discussions will provide expansion of the range to solve applied problems, improve them, and the ability to create new intellectual information technologies, based on the simulation of the elements of human thinking. | |
| 2010-2011 | Intelligent processing technology for scientific and technical information |
| Funded by Ministry of Education and Science of Ukraine (Pending funding decision) | |
| The project aims at producing methodologies, methods and tools that simplify creation, organization, search and classification of XML documents in the area of science and technology. Unlike relational data base systems that require a continuous elaboration of data base structure and software for their support, new software system will use XML database. XML format allows an effective representation of documents due to explicit definition of structure. XML database eXist stores and processes arbitrary XML documents without the need of software modifications when structure of documents changes or new document types appear. In the course of the project Web 2.0 application will be created that simplifies creation, organization and search of scientific and technical information. Additional modules for automated classification will be developed that facilitate the organization and search of XML data. Application will use open source Java Enterprise Edition Technologies (Tomcat, JBoss, Seam, GWT) and XML database eXist. | |
| 2007-2010 | Development and investigation of neural network methods for distributed representation and processing of information in information technologies |
| Funded by National Academy of Sciences of Ukraine | |
| A concept of creation of a wide class of intelligent (intellectual) information technologies based on binary sparse stochastic distributed representation of information of various type, modality, and complexity. Methods of formation of representations of complex hierarchically structured relational information using sparse codevectors, as well as properties of auto-associative memory that uses sparse distributed representations in neural network structures will be investigated. The results will be used in intelligent neural network information technologies for an efficient processing of structured numerical and textual information. | |
| 2007-2008 | Context vector representations of Ukrainian textual information for new generation of language processing information systems |
| Funded by Ministry of Education and Science of Ukraine | |
| The project aims to develop technology for the representation of semantic similarity of text fragments using the context vectors. The technology is based on methods of randomized embeddings of frequency vectors for word combinations in the chosen context. Randomized embeddings and the formation of the distributed vector representations substantially increased the computational efficiency. In the course of the project C++ libraries were created for integration with the textual information processing systems and COM Add-In for Microsoft Word was created to demonstrate technology. | |
| 2006 | Development of experimental technology for creative perception of environment by using neural technologies and imagery information models |
| Funded by Ministry of Industrial Policy of Ukraine | |
| 2004-2006 | Neural network technologies for knowledge bases |
| Funded by National Academy of Sciences of Ukraine | |
| The theoretical base of distributed representation of knowledge has been developed by conducting research of neural network methods and algorithms based on binary sparse vectors. In order to represent complex hierarchically structured information of modern knowledge bases original binding procedures that preserve information about structure in distributed representation have been proposed. Studies of distributed representations for processing of knowledge bases have been carried out on the problem of modeling analogical reasoning. It is one of the most important human mental processes that require operating structured information. To provide an interface between symbolic and distributed representations, a concept of hybrid programming language SLANG has been proposed and implemented. Language is using high-performance distributed representation processing. Both language interpreter and library were implemented in C++. | |
| 2004 | Intelligent technologies for recognition and identification of audio- and visual information based on neural network technologies |
| Funded by Ministry of Industrial Policy of Ukraine | |
| New information technology and high performance methods for recognition and identification of signals ave beeb developed. The methods for separation of audio signal mixtures (coctail party problem) improved the quality of speaker identification. The objective function for blind source separation based on minimal description length and modified matching pursuit method was proposed that is stable to additive noise in mixtures and small number of data samples. The methods for preprocessing of visual information increased the efficiency of visual recognition. | |
| 2003-2006 | Development of tools for automated information processing using analogical reasoning |
| Funded by Ministry of Education and Science of Ukraine | |
| 2003 | Development of computer technology for goal-oriented processing of textual and audio information |
| Funded by Ministry of Industrial Policy of Ukraine | |
| 2002-2006 | Novel information technologies based on highly efficient neural network methods and algorithms |
| Funded by National Academy of Sciences of Ukraine | |
| Unified distributed representations for information of various type and complexity have been developed. Those representations are used for increasing computational efficiency of machine learning and information processing methods. In order to develop and research new algorithms and technologies of information processing, modular software neurocomputer SNC has been designed and developed. It has modular and graphically configurable architecture, which is aimed to solve problems from various fields. SNC consists of WYSIWYG architecture editor and information processing blocks, which was implemented as COM modules in C++. | |
| 2002 | State Science and Technology Program "Pattern Computer": Development of technologies for semantic interpretation of textual information and semantic inter-language translation |
| Funded by Ministry of Industrial Policy of Ukraine | |
| Technology, methods, algorithms to determine semantic relatedness of words | |
| 2002 | Development of prototype for automatic speaker identification system |
| Funded by Ministry of Industrial Policy of Ukraine | |
| 2001-2003 | Development and investigation of neural network methods for modelling of cognitive processes |
| Funded by National Academy of Sciences of Ukraine | |
| 2001 | Develop methods and create tools for intellectualization of a wide range of information technologies |
| Funded by Ministry of Industrial Policy of Ukraine | |
| 2000-2001 | Creation of prototypes of new generation neurocomputers |
| Funded by Ministry of Industrial Policy of Ukraine | |
| 1999-2002 | Development of computer-aided system for automatic speaker identification and verification |
| 1998-2000 | Development of neural network technologies for information processing and their verification in the problems of decision making support, pattern recognition, and adaptive control |
| Funded by National Academy of Sciences of Ukraine | |
| 1996-1998 | High Performance Classification Based on Neural Networks with Fast Training (INTAS-93-0560) |
| Funded by INTAS | |
| 1995-1996 | Development of automated system for collection and analysis of scattered information |
| 1994-1997 | Development of methodologies and methods for knowledge representation at various hierarchical levels of neuron-like networks |
| Funded by National Academy of Sciences of Ukraine | |
| 1994-1995 | Knowledge Representation by Neural Network Structures (U4M000, U4M200) |
| Funded by International Science Foundation | |
| 1992-1994 | Neurocomputer |
| 1991-1993 | Development of methodologies and methods for solving Artificial Intelligence problems using neurocomputers |
| Funded by National Academy of Sciences of Ukraine | |
| 1991-1992 | Development of theoretical basis for creation of automated self-learning systems for processing information about object states and forecasting their behavior |
| Funded by USSR Ministry of Defence | |
| 1991 | Exploring, development and bread boarding of computational means with neural network architecture for existing and prospective flying apparatus |
| Funded by USSR Ministry of Aviation Industry | |
| 1990-1993 | Bread boarding units for technical means of artificial intelligence systems (neurocomputer) |
| Funded by USSR Ministry of Defence | |