Privacy and data security have become the new hot topic for regulators in recent years. As a resu... more Privacy and data security have become the new hot topic for regulators in recent years. As a result, Federated Learning (FL) (also called collaborative learning) has emerged as a new training paradigm that allows multiple, geographically distributed nodes to learn a Deep Learning (DL) model together without sharing their data. Blockchain is becoming a new trend as data protection and privacy are concerns in many sectors. Technology is leading the world and transforming into a global village where everything is accessible and transparent. We have presented a blockchain enabled security model using FL that can generate an enhanced DL model without sharing data and improve privacy through higher security and access rights to data. However, existing FL approaches also have unique security vulnerabilities that malicious actors can exploit and compromise the trained model. The FL method is compared to the other known approaches. Users are more likely to choose the latter option, i.e., pro...
A human brain can communicate with the outside world through another proven way that is the brain... more A human brain can communicate with the outside world through another proven way that is the brain computer interface (BCI). A BCI simply decodes the electroencephalogram (EEG) signals that are extracted from the brain activities and then sends the command to the concerned device. The device can be any interactive device like a wheelchair. The efficiency of the BCI completely depends on the process of decoding the EEG. In real world the EEG signals are polluted with different artifacts like electromyogram (EMG) and other background activity. This can cause poor decoding activity. In this paper we implement an algorithm that classifies four different motor imagery tasks where one of the body parts is involved: right hand, left hand, tongue and both feet. The electroencephalogram (EEG) data from the BCI Competition IV is used to test the three feature extraction techniques namely: time domain parameters, band power features and PCA features and with two classifiers: support vector mach...
The COVID-19 pandemic has forced much education to move into a distance learning (DL) model. The ... more The COVID-19 pandemic has forced much education to move into a distance learning (DL) model. The problem addressed in the paper is related to the increased necessity for the capacity of data, secure infrastructure, Wi-Fi possibilities, and equipment, learning resources which are needed when students connect to systems managed by institutional, national, and international organizations. Meanwhile, there have been cases when learners were not able to use technology in a secure manner, since they were requested to connect to external learning objects or systems. The research aims to develop a sustainable strategy based on a security concept model that consists of three main components: (1) security assurance; (2) users, including administration, teachers, and learners; and (3) DL organizational processes. The security concept model can be implemented at different levels of security. We modelled all the possible levels of security. To implement the security concept model, we introduce a...
This research provides analysis of existing models for competence evaluation and proposes a softw... more This research provides analysis of existing models for competence evaluation and proposes a software engineering competence remote evaluation process model. We separate competences into two domains: professional and academic. The model is developed and implemented for e-CF professional competences and academic competences evaluation for Software Engineering Master Program graduates. The remote evaluation process is based on a proposed methodology in four steps. The remote evaluation process consists of two business processes: (a) Competence test design, and then (b) Competence test completion. Business Process Model and Notation (BPMN) model was developed as a summary of both processes.
In this paper the use of convolutional neural networks (CNN) is discussed in order to solve four ... more In this paper the use of convolutional neural networks (CNN) is discussed in order to solve four class motor imagery classification problem. Analysis of viable CNN architectures and their influence on the obtained accuracy for the given task is argued. Furthermore, selection of optimal feature map image dimension, filter sizes and other CNN parameters used for network training is investigated. Methods for generating 2D feature maps from 1D feature vectors are presented for commonly used feature types. Initial results show that CNN can achieve high 68% classification accuracy for the four class motor imagery problem with less complex feature extraction techniques. It is shown that optimal accuracy highly depends on feature map dimensions, filter sizes, epoch count and other tunable factors, therefore various fine-tuning techniques must be employed. Experiments show that simple FFT energy map generation techniques are enough to reach the state-of-the-art classification accuracy for common CNN feature map sizes. This work also confirms that CNNs are able to learn a descriptive set of information needed for optimal electroencephalogram (EEG) signal classification.
Distance learning plays a significant role in solving the problem of the sustainable development ... more Distance learning plays a significant role in solving the problem of the sustainable development of society. Unfortunately, the development and implementation of distance courses are still carried out intuitively, which does not allow practical implementation of effective mathematics methods and slows down the introduction of breakthrough technologies. The aim of the research was to develop a new model and a methodology for assessing the labor costs of designing distance learning courses by university teachers based on a comprehensive assessment of the courses’ quality. The suggested model originally extends the constructive cost evaluation model already used in IT industry. The developed criteria make it possible to assess labor costs in relation to the quality of course development. The paper provides and analyzes such calculations on the example of courses in the theory of algorithms and programming.
In this paper we consider the impact of circuit realizationon the fault coverage of the test set.... more In this paper we consider the impact of circuit realizationon the fault coverage of the test set. We have performedvarious comprehensive experiments with combinationaland sequential benchmark circuits. Our experiments withcombinational circuits show that the test sets generated fora particular circuit realization fail to detect in average onlyless than one and a half percent of the stuck-at faults of there-synthesized circuit but in some cases this figure is morethan nine percent. The double test sets declined almosttwice both the maximum and the average percent ofundetected faults. The experiments exhibit that thesupplement of the test set with sensitive adjacent testpatterns significantly increases the fault coverage of the re-synthesized core. The experiments with sequential circuitsexhibit that the fault coverage’s surprisingly coincide.
Microelectronic circuits are becoming more complex and their testing is more and more complex and... more Microelectronic circuits are becoming more complex and their testing is more and more complex and expensive. If errors are detected early it is cheaper to correct them. Only high abstraction level models are available at early project stages. High level model is viewed as “black-box” in our research, not looking to its internal design. Algorithms used are based on simulation techniques and are independent of design environment. It is attempted to create test sequence, which be usable in early project stages, and later to adopt it to particular requirements. Experiments show that obtained sequences are shorter than random generated, but have similar test quality. Ill. 3, bibl. 7 (In Lithuanian; summaries in Lithuanian, English, Russian).
We investigated the influence of the random generation methods to the results of the functional d... more We investigated the influence of the random generation methods to the results of the functional delay test generation. There are the three possibilities: random generation of values 0 and 1 when the probability for the appearance of each value is 50% in every bit of test pattern; random generation of large integer values, which then are split into the sequences of 0’s and 1‘s according to the rules of the conversion of the decimal number to the binary number; anti-random generation when the presence of the earlier generated patterns is taken into account. We adopted the anti-random generation for the functional delay faults. The obtained results of the investigation indicate that the random generation has the direct influence to the results of the functional delay test generation. The largest coverage of transition fault is obtained when the anti-random generation is employed.
Lack of guidelines for implementing distance learning, lack of infrastructure, lack of competenci... more Lack of guidelines for implementing distance learning, lack of infrastructure, lack of competencies, and security-related problems were the challenges met during the pandemic. These challenges firstly fall on the administration of a higher education institution. To assist in solving the challenges of the pandemic for the administration of a higher education institution, the paper presents several models for the organization of the processes of distance learning. These models are as follows: a conceptual model of distance learning, a model of strategic planning of distance learning, a model of the assessment before the start of distance learning, a model of the preparation for distance learning, and a model of the process of distance learning and remote work. Student profile, lecturer profile, organizational environment, assessment, and planning of the infrastructure of information and communication technology (ICT), assessment and planning of the virtual learning environment, and as...
This paper introduces a blockchain-based federated learning (FL) framework with incentives for pa... more This paper introduces a blockchain-based federated learning (FL) framework with incentives for participating nodes to enhance the accuracy of classification problems. Machine learning technology has been rapidly developed and changed from a global perspective for the past few years. The FL framework is based on the Ethereum blockchain and creates an autonomous ecosystem, where nodes compete to improve the accuracy of classification problems. With privacy being one of the biggest concerns, FL makes use of the blockchain-based approach to ensure privacy and security. Another important technology that underlies the FL framework is zero-knowledge proofs (ZKPs), which ensure that data uploaded to the network are accurate and private. Basically, ZKPs allow nodes to compete fairly by only submitting accurate models to the parameter server and get rewarded for that. We have conducted an analysis and found that ZKPs can help improve the accuracy of models submitted to the parameter server an...
The motor imagery (MI) based brain-computer interface systems (BCIs) can help with new communicat... more The motor imagery (MI) based brain-computer interface systems (BCIs) can help with new communication ways. A typical electroencephalography (EEG)-based BCI system consists of several components including signal acquisition, signal pre-processing, feature extraction and feature classification. This paper focuses on the feature extraction step and proposes to use a combination of different feature extraction and feature reduction methods. The research presented in the paper explores the methods of band power, time domain parameters, fast Fourier transform and channel variance for feature extraction. These methods are investigated by combining them in pairs. The application of two feature extraction methods increases the number of selected features that can be redundant or irrelevant. The utilization of too many features can lead to wrong classification results. Therefore, the methods of feature reduction have to be applied. The following feature reduction methods are investigated: pri...
h i g h l i g h t s Hydrosodalite allows regulating the phosphogypsum hydration and setting time.... more h i g h l i g h t s Hydrosodalite allows regulating the phosphogypsum hydration and setting time. The positive effect of sonication and of hydrosodalite additive is combined. The content of radioactivity in PG does not exceed significantly average concentration. The total mass of a PG final product natural radioactivity does not create any problem.
Brain-Computer Interface (BCI) systems require intensive signal processing in order to form contr... more Brain-Computer Interface (BCI) systems require intensive signal processing in order to form control signals for electronic devices. The majority of BCI systems work by reading and interpreting cortically evoked electropotentials across the scalp via an electro encephalogram (EEG). Feature extraction and classification are the main tasks in EEG signal processing. In this paper, we propose a method to compress EEG data using discrete cosine transform (DCT). DCT takes correlated input data and concentrates its energy in just first few transform coefficients. This method is used as feature extraction step and allows reducing data size without losing important information. For classification we use feed forward artificial neural network. Experimental results show that our proposed method does not lose the important information. We conclude that the method can be successfully used for the feature extraction.
Communications in Computer and Information Science, 2013
Verification is the most crucial part of the chip design process. Test benches, which are used to... more Verification is the most crucial part of the chip design process. Test benches, which are used to test VHDL code, need perform efficiently and effectively. We present an algorithm that achieves high code coverage by analyzing the finite state machine (FSM), and control flow graph (CFG) that are constructed from the source code. The symbolic execution of VHDL (Very-high-speed integrated Hardware Description Language) code is used as well. These three elements are combined into framework (TestBenchGen) written in Python programming language and evaluated against ITC’99 benchmark suite.
Communications in Computer and Information Science, 2012
The major challenge for the semiconductor industry is to design devices in short time with comple... more The major challenge for the semiconductor industry is to design devices in short time with complex logical functionality. At the very top of the list of challenges to be solved is verification. The goal of the verification is to ensure that the design meets the logical functional requirements as defined in the logical functional specification. Verification of the devices takes 40 to 70 per cent of the total development effort for the design. The increasing complexity of hardware designs raises the need for the development of new techniques and methodologies that can provide the verification team with the means to achieve its goals quickly and with limited resources.
Privacy and data security have become the new hot topic for regulators in recent years. As a resu... more Privacy and data security have become the new hot topic for regulators in recent years. As a result, Federated Learning (FL) (also called collaborative learning) has emerged as a new training paradigm that allows multiple, geographically distributed nodes to learn a Deep Learning (DL) model together without sharing their data. Blockchain is becoming a new trend as data protection and privacy are concerns in many sectors. Technology is leading the world and transforming into a global village where everything is accessible and transparent. We have presented a blockchain enabled security model using FL that can generate an enhanced DL model without sharing data and improve privacy through higher security and access rights to data. However, existing FL approaches also have unique security vulnerabilities that malicious actors can exploit and compromise the trained model. The FL method is compared to the other known approaches. Users are more likely to choose the latter option, i.e., pro...
A human brain can communicate with the outside world through another proven way that is the brain... more A human brain can communicate with the outside world through another proven way that is the brain computer interface (BCI). A BCI simply decodes the electroencephalogram (EEG) signals that are extracted from the brain activities and then sends the command to the concerned device. The device can be any interactive device like a wheelchair. The efficiency of the BCI completely depends on the process of decoding the EEG. In real world the EEG signals are polluted with different artifacts like electromyogram (EMG) and other background activity. This can cause poor decoding activity. In this paper we implement an algorithm that classifies four different motor imagery tasks where one of the body parts is involved: right hand, left hand, tongue and both feet. The electroencephalogram (EEG) data from the BCI Competition IV is used to test the three feature extraction techniques namely: time domain parameters, band power features and PCA features and with two classifiers: support vector mach...
The COVID-19 pandemic has forced much education to move into a distance learning (DL) model. The ... more The COVID-19 pandemic has forced much education to move into a distance learning (DL) model. The problem addressed in the paper is related to the increased necessity for the capacity of data, secure infrastructure, Wi-Fi possibilities, and equipment, learning resources which are needed when students connect to systems managed by institutional, national, and international organizations. Meanwhile, there have been cases when learners were not able to use technology in a secure manner, since they were requested to connect to external learning objects or systems. The research aims to develop a sustainable strategy based on a security concept model that consists of three main components: (1) security assurance; (2) users, including administration, teachers, and learners; and (3) DL organizational processes. The security concept model can be implemented at different levels of security. We modelled all the possible levels of security. To implement the security concept model, we introduce a...
This research provides analysis of existing models for competence evaluation and proposes a softw... more This research provides analysis of existing models for competence evaluation and proposes a software engineering competence remote evaluation process model. We separate competences into two domains: professional and academic. The model is developed and implemented for e-CF professional competences and academic competences evaluation for Software Engineering Master Program graduates. The remote evaluation process is based on a proposed methodology in four steps. The remote evaluation process consists of two business processes: (a) Competence test design, and then (b) Competence test completion. Business Process Model and Notation (BPMN) model was developed as a summary of both processes.
In this paper the use of convolutional neural networks (CNN) is discussed in order to solve four ... more In this paper the use of convolutional neural networks (CNN) is discussed in order to solve four class motor imagery classification problem. Analysis of viable CNN architectures and their influence on the obtained accuracy for the given task is argued. Furthermore, selection of optimal feature map image dimension, filter sizes and other CNN parameters used for network training is investigated. Methods for generating 2D feature maps from 1D feature vectors are presented for commonly used feature types. Initial results show that CNN can achieve high 68% classification accuracy for the four class motor imagery problem with less complex feature extraction techniques. It is shown that optimal accuracy highly depends on feature map dimensions, filter sizes, epoch count and other tunable factors, therefore various fine-tuning techniques must be employed. Experiments show that simple FFT energy map generation techniques are enough to reach the state-of-the-art classification accuracy for common CNN feature map sizes. This work also confirms that CNNs are able to learn a descriptive set of information needed for optimal electroencephalogram (EEG) signal classification.
Distance learning plays a significant role in solving the problem of the sustainable development ... more Distance learning plays a significant role in solving the problem of the sustainable development of society. Unfortunately, the development and implementation of distance courses are still carried out intuitively, which does not allow practical implementation of effective mathematics methods and slows down the introduction of breakthrough technologies. The aim of the research was to develop a new model and a methodology for assessing the labor costs of designing distance learning courses by university teachers based on a comprehensive assessment of the courses’ quality. The suggested model originally extends the constructive cost evaluation model already used in IT industry. The developed criteria make it possible to assess labor costs in relation to the quality of course development. The paper provides and analyzes such calculations on the example of courses in the theory of algorithms and programming.
In this paper we consider the impact of circuit realizationon the fault coverage of the test set.... more In this paper we consider the impact of circuit realizationon the fault coverage of the test set. We have performedvarious comprehensive experiments with combinationaland sequential benchmark circuits. Our experiments withcombinational circuits show that the test sets generated fora particular circuit realization fail to detect in average onlyless than one and a half percent of the stuck-at faults of there-synthesized circuit but in some cases this figure is morethan nine percent. The double test sets declined almosttwice both the maximum and the average percent ofundetected faults. The experiments exhibit that thesupplement of the test set with sensitive adjacent testpatterns significantly increases the fault coverage of the re-synthesized core. The experiments with sequential circuitsexhibit that the fault coverage’s surprisingly coincide.
Microelectronic circuits are becoming more complex and their testing is more and more complex and... more Microelectronic circuits are becoming more complex and their testing is more and more complex and expensive. If errors are detected early it is cheaper to correct them. Only high abstraction level models are available at early project stages. High level model is viewed as “black-box” in our research, not looking to its internal design. Algorithms used are based on simulation techniques and are independent of design environment. It is attempted to create test sequence, which be usable in early project stages, and later to adopt it to particular requirements. Experiments show that obtained sequences are shorter than random generated, but have similar test quality. Ill. 3, bibl. 7 (In Lithuanian; summaries in Lithuanian, English, Russian).
We investigated the influence of the random generation methods to the results of the functional d... more We investigated the influence of the random generation methods to the results of the functional delay test generation. There are the three possibilities: random generation of values 0 and 1 when the probability for the appearance of each value is 50% in every bit of test pattern; random generation of large integer values, which then are split into the sequences of 0’s and 1‘s according to the rules of the conversion of the decimal number to the binary number; anti-random generation when the presence of the earlier generated patterns is taken into account. We adopted the anti-random generation for the functional delay faults. The obtained results of the investigation indicate that the random generation has the direct influence to the results of the functional delay test generation. The largest coverage of transition fault is obtained when the anti-random generation is employed.
Lack of guidelines for implementing distance learning, lack of infrastructure, lack of competenci... more Lack of guidelines for implementing distance learning, lack of infrastructure, lack of competencies, and security-related problems were the challenges met during the pandemic. These challenges firstly fall on the administration of a higher education institution. To assist in solving the challenges of the pandemic for the administration of a higher education institution, the paper presents several models for the organization of the processes of distance learning. These models are as follows: a conceptual model of distance learning, a model of strategic planning of distance learning, a model of the assessment before the start of distance learning, a model of the preparation for distance learning, and a model of the process of distance learning and remote work. Student profile, lecturer profile, organizational environment, assessment, and planning of the infrastructure of information and communication technology (ICT), assessment and planning of the virtual learning environment, and as...
This paper introduces a blockchain-based federated learning (FL) framework with incentives for pa... more This paper introduces a blockchain-based federated learning (FL) framework with incentives for participating nodes to enhance the accuracy of classification problems. Machine learning technology has been rapidly developed and changed from a global perspective for the past few years. The FL framework is based on the Ethereum blockchain and creates an autonomous ecosystem, where nodes compete to improve the accuracy of classification problems. With privacy being one of the biggest concerns, FL makes use of the blockchain-based approach to ensure privacy and security. Another important technology that underlies the FL framework is zero-knowledge proofs (ZKPs), which ensure that data uploaded to the network are accurate and private. Basically, ZKPs allow nodes to compete fairly by only submitting accurate models to the parameter server and get rewarded for that. We have conducted an analysis and found that ZKPs can help improve the accuracy of models submitted to the parameter server an...
The motor imagery (MI) based brain-computer interface systems (BCIs) can help with new communicat... more The motor imagery (MI) based brain-computer interface systems (BCIs) can help with new communication ways. A typical electroencephalography (EEG)-based BCI system consists of several components including signal acquisition, signal pre-processing, feature extraction and feature classification. This paper focuses on the feature extraction step and proposes to use a combination of different feature extraction and feature reduction methods. The research presented in the paper explores the methods of band power, time domain parameters, fast Fourier transform and channel variance for feature extraction. These methods are investigated by combining them in pairs. The application of two feature extraction methods increases the number of selected features that can be redundant or irrelevant. The utilization of too many features can lead to wrong classification results. Therefore, the methods of feature reduction have to be applied. The following feature reduction methods are investigated: pri...
h i g h l i g h t s Hydrosodalite allows regulating the phosphogypsum hydration and setting time.... more h i g h l i g h t s Hydrosodalite allows regulating the phosphogypsum hydration and setting time. The positive effect of sonication and of hydrosodalite additive is combined. The content of radioactivity in PG does not exceed significantly average concentration. The total mass of a PG final product natural radioactivity does not create any problem.
Brain-Computer Interface (BCI) systems require intensive signal processing in order to form contr... more Brain-Computer Interface (BCI) systems require intensive signal processing in order to form control signals for electronic devices. The majority of BCI systems work by reading and interpreting cortically evoked electropotentials across the scalp via an electro encephalogram (EEG). Feature extraction and classification are the main tasks in EEG signal processing. In this paper, we propose a method to compress EEG data using discrete cosine transform (DCT). DCT takes correlated input data and concentrates its energy in just first few transform coefficients. This method is used as feature extraction step and allows reducing data size without losing important information. For classification we use feed forward artificial neural network. Experimental results show that our proposed method does not lose the important information. We conclude that the method can be successfully used for the feature extraction.
Communications in Computer and Information Science, 2013
Verification is the most crucial part of the chip design process. Test benches, which are used to... more Verification is the most crucial part of the chip design process. Test benches, which are used to test VHDL code, need perform efficiently and effectively. We present an algorithm that achieves high code coverage by analyzing the finite state machine (FSM), and control flow graph (CFG) that are constructed from the source code. The symbolic execution of VHDL (Very-high-speed integrated Hardware Description Language) code is used as well. These three elements are combined into framework (TestBenchGen) written in Python programming language and evaluated against ITC’99 benchmark suite.
Communications in Computer and Information Science, 2012
The major challenge for the semiconductor industry is to design devices in short time with comple... more The major challenge for the semiconductor industry is to design devices in short time with complex logical functionality. At the very top of the list of challenges to be solved is verification. The goal of the verification is to ensure that the design meets the logical functional requirements as defined in the logical functional specification. Verification of the devices takes 40 to 70 per cent of the total development effort for the design. The increasing complexity of hardware designs raises the need for the development of new techniques and methodologies that can provide the verification team with the means to achieve its goals quickly and with limited resources.
Uploads
Papers by Vacius Jusas