Petro I. Kravets
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Petro I. Kravets has a total of 18 co-authors. Here are the top co-authors by number of shared publications:
- Volodymyr Shymkovych
Department of Information Systems and Technologies, National Technical University of Ukraine “Igor Sikorsky Kyiv Polytechnic Institute”, Prospect Beresteiskyi 37, 03056 Kyiv, Ukraine
6 shared publications - Lyubov Shymkovych
NTUU "KPI
4 shared publications - Anatoliy Yuchymovych Doroshenko
Institute of Software Systems NAS of Ukraine
3 shared publications Kpi
3 shared publicationsNtuu
3 shared publicationsNtuu Kpi
3 shared publications
Latest Publications
- Open Access
- 26 December 2024
- Published by Igor Sikorsky Kyiv Polytechnic Institute in Information, Computing and Intelligent systems
- No. 5,p. 150-158
Scientific article summarization model with unbounded input length
In recent years, the exponential growth of scientific literature has made it increasingly difficult for researchers and practitioners to keep up with new discoveries and developments in their fields. Thanks to this, text summarization has become one of the primary tasks of natural language processing. Abstractive summarization of long documents, such as scientific articles, requires large neural networks with high memory and computation requirements. Therefore, it is all the more important to find ways to increase the efficiency of long document summarization models. The objects of this research are long document summarization transformer models and the Unlimiformer cross-attention modification. The article reviews the basic principles of transformer attention, which constitutes the primary computational expense in transformer models. More efficient self-attention approaches used for long document summarization models are described, such as the global+sliding window attention used by Longformer. The cross-attention mechanism of Unlimiformer, which allows a model to have unbounded input length, is described in detail. The objective of the study is the development and evaluation of a long document summarization model using the Unlimiformer modification. To achieve this goal, a Longformer Decoder-Encoder model pretrained on the arXiv dataset is modified with Unlimiformer cross-attention. This modification can be applied without additional model fine-tuning, avoiding the cost of further training a large sequence length model. The developed model was evaluated on the arXiv dataset using the ROUGE-1, ROUGE-2 and ROUGE-L metrics. The developed model showed improved results compared to the baseline model, demonstrating the viability of using this approach to improve long document summarization models.
- 26 December 2024
- Published by Igor Sikorsky Kyiv Polytechnic Institute in Information, Computing and Intelligent systems
- No. 5,p. 150-158
Open AccessScientific article summarization model with unbounded input length
In recent years, the exponential growth of scientific literature has made it increasingly difficult for researchers and practitioners to keep up with new discoveries and developments in their fields. Thanks to this, text summarization has become one of the primary tasks of natural language processing. Abstractive summarization of long documents, such as scientific articles, requires large neural networks with high memory and computation requirements. Therefore, it is all the more important to find ways to increase the efficiency of long document summarization models. The objects of this research are long document summarization transformer models and the Unlimiformer cross-attention modification. The article reviews the basic principles of transformer attention, which constitutes the primary computational expense in transformer models. More efficient self-attention approaches used for long document summarization models are described, such as the global+sliding window attention used by Longformer. The cross-attention mechanism of Unlimiformer, which allows a model to have unbounded input length, is described in detail. The objective of the study is the development and evaluation of a long document summarization model using the Unlimiformer modification. To achieve this goal, a Longformer Decoder-Encoder model pretrained on the arXiv dataset is modified with Unlimiformer cross-attention. This modification can be applied without additional model fine-tuning, avoiding the cost of further training a large sequence length model. The developed model was evaluated on the arXiv dataset using the ROUGE-1, ROUGE-2 and ROUGE-L metrics. The developed model showed improved results compared to the baseline model, demonstrating the viability of using this approach to improve long document summarization models.
- Open Access
- 1 January 2024
- Published by National Academy of Sciences of Ukraine (Co. LTD Ukrinformnauka) (Publications) in PROBLEMS IN PROGRAMMING
- No. 1,p. 23-29
An intelligent chatbot for evaluating the emotional colouring of a message and responding accordingly
A recurrent neural network model, a database designed for neural network training, and a software tool for interacting with a bot have all been created. A large dataset (50 thousand comments) containing different reviews and sentiments was collected and annotated to successfully train and validate the model. It was also translated into Ukrainian language with the help of an automatic translator. The architecture of the neural network model underwent optimization to enhance classification outcomes. Furthermore, work was conducted on enhancing the user interface. The developed application was tested, and the results were demonstrated. The resulting model demonstrated accuracy 85% in determining sentiments. The implemented application has got basic design (which can be customized) and some settings for chatbot. Further improvement of the model’s classification quality can be achieved by collecting a larger and better organised dataset or by researching other RNN architectures.
- 1 January 2024
- Published by National Academy of Sciences of Ukraine (Co. LTD Ukrinformnauka) (Publications) in PROBLEMS IN PROGRAMMING
- No. 1,p. 23-29
Open AccessAn intelligent chatbot for evaluating the emotional colouring of a message and responding accordingly
A recurrent neural network model, a database designed for neural network training, and a software tool for interacting with a bot have all been created. A large dataset (50 thousand comments) containing different reviews and sentiments was collected and annotated to successfully train and validate the model. It was also translated into Ukrainian language with the help of an automatic translator. The architecture of the neural network model underwent optimization to enhance classification outcomes. Furthermore, work was conducted on enhancing the user interface. The developed application was tested, and the results were demonstrated. The resulting model demonstrated accuracy 85% in determining sentiments. The implemented application has got basic design (which can be customized) and some settings for chatbot. Further improvement of the model’s classification quality can be achieved by collecting a larger and better organised dataset or by researching other RNN architectures.
- Open Access
- 1 June 2023
- Published by National Academy of Sciences of Ukraine (Co. LTD Ukrinformnauka) (Publications) in PROBLEMS IN PROGRAMMING
- No. 2,p. 10-23
Software system for laser targeting dropped ammunition
For last few years usage of small commercial unmanned aerial vehicles (UAV) on battlefield was drastically increased. In addition to use UAVs as spotter, they more and more frequently applied to drop small ammunition on enemy forces. In this paper laser targeting system that suitable for use on lightweight, mobile drones that drop freefalling ammunition was developed. Laser targeting systems are used for detecting and tracking the targets in the battlefield by obtaining the position information of these targets which is being designated and illuminated by UAV’s laser designator then the laser light reflected from these targets and come to the input of the optical systems of the dropped ammunition. An optical system for collimating the reflected beam from the target is one of the main components of the laser targeting system (LTS). The Quad Detector (QD) as a part of this optical system is simply consists of four photodiodes capable of detecting light spot projected on its surfaces and determine the deviation position of the laser spot from its center. It converts the incident laser spot to its corresponding photocurrent, the readout circuit that filter and convert the photocurrent to its corresponding voltage and the tracking system that is controlling the laser seeker movement to track the intended target based on the feedback information of the QD depending on the real position of the tracking platform. Developed system will ensure that munitions are accurately hit the target, reducing unnecessary casualties and damage, as well as reducing the number of sorties required and the time spent in hazardous airspace for the unmanned aerial vehicle.
- 1 June 2023
- Published by National Academy of Sciences of Ukraine (Co. LTD Ukrinformnauka) (Publications) in PROBLEMS IN PROGRAMMING
- No. 2,p. 10-23
Open AccessSoftware system for laser targeting dropped ammunition
For last few years usage of small commercial unmanned aerial vehicles (UAV) on battlefield was drastically increased. In addition to use UAVs as spotter, they more and more frequently applied to drop small ammunition on enemy forces. In this paper laser targeting system that suitable for use on lightweight, mobile drones that drop freefalling ammunition was developed. Laser targeting systems are used for detecting and tracking the targets in the battlefield by obtaining the position information of these targets which is being designated and illuminated by UAV’s laser designator then the laser light reflected from these targets and come to the input of the optical systems of the dropped ammunition. An optical system for collimating the reflected beam from the target is one of the main components of the laser targeting system (LTS). The Quad Detector (QD) as a part of this optical system is simply consists of four photodiodes capable of detecting light spot projected on its surfaces and determine the deviation position of the laser spot from its center. It converts the incident laser spot to its corresponding photocurrent, the readout circuit that filter and convert the photocurrent to its corresponding voltage and the tracking system that is controlling the laser seeker movement to track the intended target based on the feedback information of the QD depending on the real position of the tracking platform. Developed system will ensure that munitions are accurately hit the target, reducing unnecessary casualties and damage, as well as reducing the number of sorties required and the time spent in hazardous airspace for the unmanned aerial vehicle.
- Open Access
- 1 June 2023
- Published by National Academy of Sciences of Ukraine (Co. LTD Ukrinformnauka) (Publications) in PROBLEMS IN PROGRAMMING
- No. 2,p. 59-66
A Convolutional Neural Network Model and Software Tool for Classifying the Presence of a Medical Mask on a Human Face
A model of a convolutional neural network, a database for training a neural network, and a software tool for classifying the presence of a medical mask on a person’s face, which allows recognizing the presence of a medical mask from the transmitted image, have been developed. The structure of the neural network model was optimized to improve classification results. In addition, the development of the user interface was carried out. The developed application was tested on a set of random images. The resulting model demonstrated high accuracy and robustness in solving the task of classifying the presence of a medical mask on a person’s face, which allows automating measures to protect people from the spread of diseases. The implemented application meets the requirements for speed and quality of classification. Further improvement of the classification quality of CNN can be done by collecting a larger dataset and researching other CNN architectures.
- 1 June 2023
- Published by National Academy of Sciences of Ukraine (Co. LTD Ukrinformnauka) (Publications) in PROBLEMS IN PROGRAMMING
- No. 2,p. 59-66
Open AccessA Convolutional Neural Network Model and Software Tool for Classifying the Presence of a Medical Mask on a Human Face
A model of a convolutional neural network, a database for training a neural network, and a software tool for classifying the presence of a medical mask on a person’s face, which allows recognizing the presence of a medical mask from the transmitted image, have been developed. The structure of the neural network model was optimized to improve classification results. In addition, the development of the user interface was carried out. The developed application was tested on a set of random images. The resulting model demonstrated high accuracy and robustness in solving the task of classifying the presence of a medical mask on a person’s face, which allows automating measures to protect people from the spread of diseases. The implemented application meets the requirements for speed and quality of classification. Further improvement of the classification quality of CNN can be done by collecting a larger dataset and researching other CNN architectures.
- Open Access
- 13 March 2021
- Published by Springer Nature in Neural Computing and Applications
- Vol. 33 (15) , 9467-9479
Hardware implementation of radial-basis neural networks with Gaussian activation functions on FPGA
This article introduces a method for realizing the Gaussian activation function of radial-basis (RBF) neural networks with their hardware implementation on field-programmable gaits area (FPGAs). The results of modeling of the Gaussian function on FPGA chips of different families have been presented. RBF neural networks of various topologies have been synthesized and investigated. The hardware component implemented by this algorithm is an RBF neural network with four neurons of the latent layer and one neuron with a sigmoid activation function on an FPGA using 16-bit numbers with a fixed point, which took 1193 logic matrix gate (LUTs—LookUpTable). Each hidden layer neuron of the RBF network is designed on an FPGA as a separate computing unit. The speed as a total delay of the combination scheme of the block RBF network was 101.579 ns. The implementation of the Gaussian activation functions of the hidden layer of the RBF network occupies 106 LUTs, and the speed of the Gaussian activation functions is 29.33 ns. The absolute error is ± 0.005. The Spartan 3 family of chips for modeling has been used to get these results. Modeling on chips of other series has been also introduced in the article. RBF neural networks of various topologies have been synthesized and investigated. Hardware implementation of RBF neural networks with such speed allows them to be used in real-time control systems for high-speed objects.
- 13 March 2021
- Published by Springer Nature in Neural Computing and Applications
- Vol. 33 (15) , 9467-9479
Open AccessHardware implementation of radial-basis neural networks with Gaussian activation functions on FPGA
This article introduces a method for realizing the Gaussian activation function of radial-basis (RBF) neural networks with their hardware implementation on field-programmable gaits area (FPGAs). The results of modeling of the Gaussian function on FPGA chips of different families have been presented. RBF neural networks of various topologies have been synthesized and investigated. The hardware component implemented by this algorithm is an RBF neural network with four neurons of the latent layer and one neuron with a sigmoid activation function on an FPGA using 16-bit numbers with a fixed point, which took 1193 logic matrix gate (LUTs—LookUpTable). Each hidden layer neuron of the RBF network is designed on an FPGA as a separate computing unit. The speed as a total delay of the combination scheme of the block RBF network was 101.579 ns. The implementation of the Gaussian activation functions of the hidden layer of the RBF network occupies 106 LUTs, and the speed of the Gaussian activation functions is 29.33 ns. The absolute error is ± 0.005. The Spartan 3 family of chips for modeling has been used to get these results. Modeling on chips of other series has been also introduced in the article. RBF neural networks of various topologies have been synthesized and investigated. Hardware implementation of RBF neural networks with such speed allows them to be used in real-time control systems for high-speed objects.
- 1 September 2017
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 2, 947-951
Method and technology of synthesis of neural network models of object control with their hardware implementation on FPGA
This work presents a method and technology of synthesis of direct and inverse neural network models of object control, and their hardware implementation of programmable logic integrated circuits, that will automate the process of synthesis of such models and shorten the time of development and research.
- 1 September 2017
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 2, 947-951
Method and technology of synthesis of neural network models of object control with their hardware implementation on FPGA
This work presents a method and technology of synthesis of direct and inverse neural network models of object control, and their hardware implementation of programmable logic integrated circuits, that will automate the process of synthesis of such models and shorten the time of development and research.
- 1 January 2011
- Published by Begell House Inc. Publishers in Journal of Long-Term Effects of Medical Implants
- Vol. 43 (4) , 70-77
Methods of Hardware and Software Realization of Adaptive Neural Network PID Controller on FPGA-Chip
The methods of hardware and software implementation of adaptive neural network PID controller on FPGA-chip is considered, the stepwise algorithm of such controller synthesis is presented. The model of the controller is presented in the System Generator for DSP, calculations of optimal bit network of data which ensures correct operation of the PID controller are performed.
- 1 January 2011
- Published by Begell House Inc. Publishers in Journal of Long-Term Effects of Medical Implants
- Vol. 43 (4) , 70-77
Methods of Hardware and Software Realization of Adaptive Neural Network PID Controller on FPGA-Chip
The methods of hardware and software implementation of adaptive neural network PID controller on FPGA-chip is considered, the stepwise algorithm of such controller synthesis is presented. The model of the controller is presented in the System Generator for DSP, calculations of optimal bit network of data which ensures correct operation of the PID controller are performed.