Pub Date : 2019-04-12DOI: 10.15622/SP.18.2.354-389
Y. Seliverstov, Viktoriya Chigur, Arseniy Sazanov, S. Seliverstov, A. Svistunova
As a result of the analysis, it was revealed that social networks (Vkontakte, Facebook), thematic communities in microblogging networks (Twitter), resources for travelers (TripAdvisor), transport portals (Autostrada) are a source of up-to-date and operational information about the traffic situation, the quality of transport services and passenger satisfaction with the quality of levels of transport services. However, the existing transport monitoring systems do not contain software tools capable of collecting and analyzing traffic information located in the Internet environment. This paper discusses the task of building a system for automatically retrieving and classifying road traffic information from transport Internet portals and testing the developed system for analyzing the transport networks of Crimea and the city of Sevastopol. To solve this problem, an analysis of open source libraries for thematic data collection and analysis was carried out. An algorithm for extracting and analyzing texts has been developed. A crawler was developed using the Scrapy package in Python3, and user feedback from the portal http://autostrada.info/ru was collected on the state of the transport system of Crimea and the city of Sevastopol. For texts lemmatization and vector text transformation, the tf, idf, tf-idf methods and their implementation in the Scikit-Learn library were considered: CountVectorizer and TF-IDF Vectorizer. For word processing, Bag-of-Words and n-gram methods were considered. During the development of the classifier model, the naive Bayes algorithm (MultinomialNB) and the linear classifier model with optimization of the stochastic gradient descent (SGDClassifier) were used. As a training sample, a corpus of 225,000 labeled texts from the Twitter resource was used. The classifier was trained, during which the cross-validation strategy and the ShuffleSplit method were used. Testing and comparison of the results of the pitch classification were carried out. According to the results of validation, the linear model with the n-gram scheme [1, 3] and the vectorizer TF-IDF turned out to be the best. During the approbation of the developed system, the collection and analysis of reviews related to the quality of transport networks of the Republic of Crimea and the city of Sevastopol were conducted. Conclusions are drawn and prospects for further functional development of the developed tools are defined.
{"title":"Sentiment Analysis of \"AUTOSTRADA.INFO/RU\" Users’ Comments","authors":"Y. Seliverstov, Viktoriya Chigur, Arseniy Sazanov, S. Seliverstov, A. Svistunova","doi":"10.15622/SP.18.2.354-389","DOIUrl":"https://doi.org/10.15622/SP.18.2.354-389","url":null,"abstract":"As a result of the analysis, it was revealed that social networks (Vkontakte, Facebook), thematic communities in microblogging networks (Twitter), resources for travelers (TripAdvisor), transport portals (Autostrada) are a source of up-to-date and operational information about the traffic situation, the quality of transport services and passenger satisfaction with the quality of levels of transport services. However, the existing transport monitoring systems do not contain software tools capable of collecting and analyzing traffic information located in the Internet environment. This paper discusses the task of building a system for automatically retrieving and classifying road traffic information from transport Internet portals and testing the developed system for analyzing the transport networks of Crimea and the city of Sevastopol. To solve this problem, an analysis of open source libraries for thematic data collection and analysis was carried out. An algorithm for extracting and analyzing texts has been developed. A crawler was developed using the Scrapy package in Python3, and user feedback from the portal http://autostrada.info/ru was collected on the state of the transport system of Crimea and the city of Sevastopol. For texts lemmatization and vector text transformation, the tf, idf, tf-idf methods and their implementation in the Scikit-Learn library were considered: CountVectorizer and TF-IDF Vectorizer. For word processing, Bag-of-Words and n-gram methods were considered. During the development of the classifier model, the naive Bayes algorithm (MultinomialNB) and the linear classifier model with optimization of the stochastic gradient descent (SGDClassifier) were used. As a training sample, a corpus of 225,000 labeled texts from the Twitter resource was used. The classifier was trained, during which the cross-validation strategy and the ShuffleSplit method were used. Testing and comparison of the results of the pitch classification were carried out. According to the results of validation, the linear model with the n-gram scheme [1, 3] and the vectorizer TF-IDF turned out to be the best. During the approbation of the developed system, the collection and analysis of reviews related to the quality of transport networks of the Republic of Crimea and the city of Sevastopol were conducted. Conclusions are drawn and prospects for further functional development of the developed tools are defined.","PeriodicalId":53447,"journal":{"name":"SPIIRAS Proceedings","volume":"24 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2019-04-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"83216766","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-02-21DOI: 10.15622/SP.18.1.202-229
V. Mironov, Y. Mironov, I. Fominov
The article is devoted to the development of algorithms for calculating energy-optimal programs for controlling the approach of a spacecraft to an orbital object at the long-range guidance stage using the maximum principle L.S. Pontryagin. It is assumed that the spacecraft is equipped with a longitudinal propulsion system running on chemical fuel. The programs of optimal change in the second fuel consumption and the vector of cosine guides, which determine the orientation of the thrust force of the propulsion system, should be determined. As a criterion for optimal control, we consider a functional that determines the minimum consumption of the working fluid. The optimal control problem is solved in a limited region of the state space, which is determined by the range of variation of the angular range of the spacecraft within one revolution. The full equations of the boundary value problem of the maximum principle are given using the model of the object's motion in the normal gravitational field of the Earth, the boundary conditions, and also the analytical dependencies that determine the control structure in the optimal mode. The boundary optimization problem is solved with the Newton method. To determine the initial approximation of conjugate variables, an analytical solution is given to the problem of energy-optimal approach control in a uniform central field. The results of numerical studies of energy-optimal programs to control the interception in the normal gravitational field of the Earth with the final pitch at the stage of long-range guidance are presented. In general, the use of optimal approach control algorithms in the airborne and ground complex allows to reduce fuel costs when performing the maneuver, reduce time, and expand the reachability area of the target objects. In addition, the optimal solution can be considered as a reference, with which it is necessary to compare various variants of approximate control algorithms, evaluate their quality and make informed decisions on their practical use.
{"title":"Energetically Optimal Control of the Convergence of the Spacecraft in Non-Central Gravitational Field of the Earth on the far Guidance Stage","authors":"V. Mironov, Y. Mironov, I. Fominov","doi":"10.15622/SP.18.1.202-229","DOIUrl":"https://doi.org/10.15622/SP.18.1.202-229","url":null,"abstract":"The article is devoted to the development of algorithms for calculating energy-optimal programs for controlling the approach of a spacecraft to an orbital object at the long-range guidance stage using the maximum principle L.S. Pontryagin. It is assumed that the spacecraft is equipped with a longitudinal propulsion system running on chemical fuel. The programs of optimal change in the second fuel consumption and the vector of cosine guides, which determine the orientation of the thrust force of the propulsion system, should be determined. As a criterion for optimal control, we consider a functional that determines the minimum consumption of the working fluid. The optimal control problem is solved in a limited region of the state space, which is determined by the range of variation of the angular range of the spacecraft within one revolution. The full equations of the boundary value problem of the maximum principle are given using the model of the object's motion in the normal gravitational field of the Earth, the boundary conditions, and also the analytical dependencies that determine the control structure in the optimal mode. The boundary optimization problem is solved with the Newton method. To determine the initial approximation of conjugate variables, an analytical solution is given to the problem of energy-optimal approach control in a uniform central field. The results of numerical studies of energy-optimal programs to control the interception in the normal gravitational field of the Earth with the final pitch at the stage of long-range guidance are presented. In general, the use of optimal approach control algorithms in the airborne and ground complex allows to reduce fuel costs when performing the maneuver, reduce time, and expand the reachability area of the target objects. In addition, the optimal solution can be considered as a reference, with which it is necessary to compare various variants of approximate control algorithms, evaluate their quality and make informed decisions on their practical use.","PeriodicalId":53447,"journal":{"name":"SPIIRAS Proceedings","volume":"43 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2019-02-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"83697432","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-02-21DOI: 10.15622/SP.18.1.123-147
V. Petrenko, F. Tebueva, M. Gurchinsky, V. Antonov, A. Pavlov
The most important task of modern robotics is the development of robots to perform the work in potentially dangerous fields which can cause the risk to human health. Currently robotic systems can not become a full replacement for man for solving complex problems in a dynamic environment despite an active development of artificial intelligence technologies. The robots that implement the copying type of control or the so-called virtual presence of the operator are the most advanced for use in the nearest future. The principle of copying control is based on the motion capture of the remote operator and the formation of control signals for the robot’s drives. A tracking system or systems based on movement planning can be used to control the drives. The tracking systems are simpler, but systems based on motion planning allow to achieve more smooth motion and less wear on the parts of the control object. An artificial delay between the movements of the operator and the control object for necessary data collection is used to implement the control-based motion planning. The aim of research is a reduction of delay, which appears when controlling the anthropomorphic manipulator drives based on the solution of the inverse dynamic problem, when real time copying type of control is used . For motion path planning it is proposed to use forecast values of the generalized coordinates for manipulator. Based on the measured values of the generalized coordinates of the operator's hand, time series are formed and their prediction is performed. Predictive values of generalized coordinates are used in planning the anthropomorphic manipulator trajectory and solving the inverse dynamic problem. Prediction is based on linear regression with relatively low computational complexity, which is an important criterion for the system operation in the real time operation mode. The developed mathematical apparatus, based on prediction parameters and maximum permissible accelerations of the manipulator drives, allows to find a theoretical estimate of error values limits for planning the operator's hand trajectory using the proposed approach for specific tasks. The adequacy of the maximum theoretical value of the prediction error, as well as the prospects of the proposed approach for testing in practice is confirmed by the software simulation in Matlab environment.
{"title":"Predictive Assessment of Operator's Hand Trajectory with the Copying Type of Control for Solution of the Inverse Dynamic Problem","authors":"V. Petrenko, F. Tebueva, M. Gurchinsky, V. Antonov, A. Pavlov","doi":"10.15622/SP.18.1.123-147","DOIUrl":"https://doi.org/10.15622/SP.18.1.123-147","url":null,"abstract":"The most important task of modern robotics is the development of robots to perform the work in potentially dangerous fields which can cause the risk to human health. Currently robotic systems can not become a full replacement for man for solving complex problems in a dynamic environment despite an active development of artificial intelligence technologies. \u0000The robots that implement the copying type of control or the so-called virtual presence of the operator are the most advanced for use in the nearest future. The principle of copying control is based on the motion capture of the remote operator and the formation of control signals for the robot’s drives. A tracking system or systems based on movement planning can be used to control the drives. The tracking systems are simpler, but systems based on motion planning allow to achieve more smooth motion and less wear on the parts of the control object. An artificial delay between the movements of the operator and the control object for necessary data collection is used to implement the control-based motion planning. \u0000The aim of research is a reduction of delay, which appears when controlling the anthropomorphic manipulator drives based on the solution of the inverse dynamic problem, when real time copying type of control is used . For motion path planning it is proposed to use forecast values of the generalized coordinates for manipulator. Based on the measured values of the generalized coordinates of the operator's hand, time series are formed and their prediction is performed. Predictive values of generalized coordinates are used in planning the anthropomorphic manipulator trajectory and solving the inverse dynamic problem. Prediction is based on linear regression with relatively low computational complexity, which is an important criterion for the system operation in the real time operation mode. The developed mathematical apparatus, based on prediction parameters and maximum permissible accelerations of the manipulator drives, allows to find a theoretical estimate of error values limits for planning the operator's hand trajectory using the proposed approach for specific tasks. The adequacy of the maximum theoretical value of the prediction error, as well as the prospects of the proposed approach for testing in practice is confirmed by the software simulation in Matlab environment.","PeriodicalId":53447,"journal":{"name":"SPIIRAS Proceedings","volume":"25 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2019-02-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"82987130","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Lavrenov Lavrenov, E. Magid, M. Fumitoshi, M. Svinin, J. Suthakorn
Path planning for autonomous mobile robots is an important task within robotics field. It is common to use one of the two classical approaches in path planning: a global approach when an entire map of a working environment is available for a robot or local methods, which require the robot to detect obstacles with a variety of onboard sensors as the robot traverses the environment. In our previous work, a multi-criteria spline algorithm prototype for a global path construction was developed and tested in Matlab environment. The algorithm used the Voronoi graph for computing an initial path that serves as a starting point of the iterative method. This approach allowed finding a path in all map configurations whenever the path existed. During the iterative search, a cost function with a number of different criteria and associated weights was guiding further path optimization. A potential field method was used to implement some of the criteria. This paper describes an implementation of a modified spline-based algorithm that could be used with real autonomous mobile robots. Equations of the characteristic criteria of a path optimality were further modified. The obstacle map was previously presented as intersections of a finite number of circles with various radii. However, in real world environments, obstacles’ data is a dynamically changing probability map that could be based on an occupancy grid. Moreover, the robot is no longer a geometric point. To implement the spline algorithm and further use it with real robots, the source code of the Matlab environment prototype was transferred into C++ programming language. The testing of the method and the multi criteria cost function optimality was carried out in ROS/Gazebo environment, which recently has become a standard for programming and modeling robotic devices and algorithms. The resulting spline-based path planning algorithm could be used on any real robot, which is equipped with a laser rangefinder. The algorithm operates in real time and the influence of the objective function criteria parameters are available for dynamic tuning during a robot motion.
{"title":"Development and Implementation of Spline-based Path Planning Algorithm in ROS/Gazebo Environment","authors":"Lavrenov Lavrenov, E. Magid, M. Fumitoshi, M. Svinin, J. Suthakorn","doi":"10.15622/SP.18.1.57-84","DOIUrl":"https://doi.org/10.15622/SP.18.1.57-84","url":null,"abstract":"Path planning for autonomous mobile robots is an important task within robotics field. It is common to use one of the two classical approaches in path planning: a global approach when an entire map of a working environment is available for a robot or local methods, which require the robot to detect obstacles with a variety of onboard sensors as the robot traverses the environment. \u0000In our previous work, a multi-criteria spline algorithm prototype for a global path construction was developed and tested in Matlab environment. The algorithm used the Voronoi graph for computing an initial path that serves as a starting point of the iterative method. This approach allowed finding a path in all map configurations whenever the path existed. During the iterative search, a cost function with a number of different criteria and associated weights was guiding further path optimization. A potential field method was used to implement some of the criteria. \u0000This paper describes an implementation of a modified spline-based algorithm that could be used with real autonomous mobile robots. Equations of the characteristic criteria of a path optimality were further modified. The obstacle map was previously presented as intersections of a finite number of circles with various radii. However, in real world environments, obstacles’ data is a dynamically changing probability map that could be based on an occupancy grid. Moreover, the robot is no longer a geometric point. \u0000To implement the spline algorithm and further use it with real robots, the source code of the Matlab environment prototype was transferred into C++ programming language. The testing of the method and the multi criteria cost function optimality was carried out in ROS/Gazebo environment, which recently has become a standard for programming and modeling robotic devices and algorithms. \u0000The resulting spline-based path planning algorithm could be used on any real robot, which is equipped with a laser rangefinder. The algorithm operates in real time and the influence of the objective function criteria parameters are available for dynamic tuning during a robot motion.","PeriodicalId":53447,"journal":{"name":"SPIIRAS Proceedings","volume":"60 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2019-02-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"77216435","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-02-18DOI: 10.15622/SP.18.1.230-262
L. Babenko, A. Trepacheva
The security of two recently proposed symmetric homomorphic encryption schemes based on residue system is analyzed. Both schemes have a high computational efficiency since using residue system naturally allows parallelizing computations. So they could be good candidates to protect the data in clouds. But to the best of our knowledge there is a lack of security analysis for these encryption schemes. It should be noted that the first cryptosystem under our consideration was already considered in literature. The sketch of adaptive chosen-plaintext attack was proposed and estimation of its success was given. In this paper the attack is analyzed and it is shown that in some cases it may work incorrectly. Also more general algorithm of known-plaintext attack is presented. Theoretical estimations of probability to recover the key using it and practical estimations of this probability obtained during the experiments are provided. The security of the second cryptosystem has not been analyzed yet and we fill this gap for known-plaintext attack. The dependency between the number of «plaintext, ciphertext» pairs required to recover the key and parameters of the cryptosystem is analyzed. Also some recommendations for increasing the security level are provided. The final conclusion of our analysis is that both cryptosystems are vulnerable to known-plaintext attack. And it may be dangerous to encrypt private data using them. Finally it should be noted that the key element of the proposed attacks is the algorithm of computing the greatest common divisor. So their computational complexity depends polynomially on the size of input data.
{"title":"Towards Unsecurity of Two Homomorphic Encryptions Based on Residue System","authors":"L. Babenko, A. Trepacheva","doi":"10.15622/SP.18.1.230-262","DOIUrl":"https://doi.org/10.15622/SP.18.1.230-262","url":null,"abstract":"The security of two recently proposed symmetric homomorphic encryption schemes based on residue system is analyzed. \u0000Both schemes have a high computational efficiency since using residue system naturally allows parallelizing computations. So they could be good candidates to protect the data in clouds. But to the best of our knowledge there is a lack of security analysis for these encryption schemes. \u0000It should be noted that the first cryptosystem under our consideration was already considered in literature. \u0000The sketch of adaptive chosen-plaintext attack was proposed and estimation of its success was given. \u0000In this paper the attack is analyzed and it is shown that in some cases it may work incorrectly. Also more general algorithm of known-plaintext attack is presented. Theoretical estimations of probability to recover the key using it and practical estimations of this probability obtained during the experiments are provided. \u0000The security of the second cryptosystem has not been analyzed yet and we fill this gap for known-plaintext attack. The dependency between the number of «plaintext, ciphertext» pairs required to recover the key and parameters of the cryptosystem is analyzed. Also some recommendations for increasing the security level are provided. \u0000The final conclusion of our analysis is that both cryptosystems are vulnerable to known-plaintext attack. And it may be dangerous to encrypt private data using them. \u0000Finally it should be noted that the key element of the proposed attacks is the algorithm of computing the greatest common divisor. So their computational complexity depends polynomially on the size of input data.","PeriodicalId":53447,"journal":{"name":"SPIIRAS Proceedings","volume":"70 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2019-02-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"84457967","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
A. Gorobtsov, A. Andreev, A. Markov, A. Skorikov, P. Tarasov
The problem of walking robots controlled motion synthesis by the inverse dynamic method is considered. The inverse dynamic method equations are represented by the methods of multibody system dynamics as free bodies motion equations and constraint equations. The variety of constraint equations group are introduced to specify the robot gait, to implement the robot stability conditions and to coordinate specified robot links movement. The key feature of the inverse dynamic method equations in this formulation is the presence of the second derivatives of the system coordinates in the constraint equations expressing the stability conditions that ensure the maintenance of the vertical position by the robot. The determined solution of such equations in general case is impossible due to the uncertainty of the initial conditions for the Lagrange multipliers. An approximate method for solving the inverse dynamic without taking into account the inertial components in the constraint equations that determine the stability of the robot is considered. Constraint equations that determine the coordinate movement of individual robot links and required for unique problem solving based on approximate equations are presented. The implementation of program motion synthesis methods in the control system of the humanoid robot AR-600 is presented. The comparison of theoretical and experimental parameters of controlled motion is performed. It has been established that with the achieved high accuracy of the robot links tracking drives control with an error of several percent, the indicators of the robot's absolute movements, in particular, the angles of roll, yaw and pitch, differ from the programmed by 30-40%. It’s shown that proposed method allows to synthesize robot control in quasistatic mode for different movement types such as moving forward, sideways, walking on stairs, inclinations etc.
{"title":"Features of solving the inverse dynamic method equations for the synthesis of stable walking robots controlled motion","authors":"A. Gorobtsov, A. Andreev, A. Markov, A. Skorikov, P. Tarasov","doi":"10.15622/SP.18.1.85-122","DOIUrl":"https://doi.org/10.15622/SP.18.1.85-122","url":null,"abstract":"The problem of walking robots controlled motion synthesis by the inverse dynamic method is considered. The inverse dynamic method equations are represented by the methods of multibody system dynamics as free bodies motion equations and constraint equations. The variety of constraint equations group are introduced to specify the robot gait, to implement the robot stability conditions and to coordinate specified robot links movement. The key feature of the inverse dynamic method equations in this formulation is the presence of the second derivatives of the system coordinates in the constraint equations expressing the stability conditions that ensure the maintenance of the vertical position by the robot. The determined solution of such equations in general case is impossible due to the uncertainty of the initial conditions for the Lagrange multipliers. An approximate method for solving the inverse dynamic without taking into account the inertial components in the constraint equations that determine the stability of the robot is considered. Constraint equations that determine the coordinate movement of individual robot links and required for unique problem solving based on approximate equations are presented. The implementation of program motion synthesis methods in the control system of the humanoid robot AR-600 is presented. The comparison of theoretical and experimental parameters of controlled motion is performed. It has been established that with the achieved high accuracy of the robot links tracking drives control with an error of several percent, the indicators of the robot's absolute movements, in particular, the angles of roll, yaw and pitch, differ from the programmed by 30-40%. It’s shown that proposed method allows to synthesize robot control in quasistatic mode for different movement types such as moving forward, sideways, walking on stairs, inclinations etc.","PeriodicalId":53447,"journal":{"name":"SPIIRAS Proceedings","volume":"21 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2019-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"77853330","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Recently, Speech Emotion Recognition (SER) has become an important research topic of affective computing. It is a difficult problem, where some of the greatest challenges lie in the feature selection and representation tasks. A good feature representation should be able to reflect global trends as well as temporal structure of the signal, since emotions naturally evolve in time; it has become possible with the advent of Recurrent Neural Networks (RNN), which are actively used today for various sequence modeling tasks. This paper proposes a hybrid approach to feature representation, which combines traditionally engineered statistical features with Long Short-Term Memory (LSTM) sequence representation in order to take advantage of both short-term and long-term acoustic characteristics of the signal, therefore capturing not only the general trends but also temporal structure of the signal. The evaluation of the proposed method is done on three publicly available acted emotional speech corpora in three different languages, namely RUSLANA (Russian speech), BUEMODB (Turkish speech) and EMODB (German speech). Compared to the traditional approach, the results of our experiments show an absolute improvement of 2.3% and 2.8% for two out of three databases, and a comparative performance on the third. Therefore, provided enough training data, the proposed method proves effective in modelling emotional content of speech utterances.
{"title":"Modeling Short-Term and Long-Term Dependencies of the Speech Signal for Paralinguistic Emotion Classification","authors":"Oxana Verkholyak, Heysem Kaya, Alexey Karpov","doi":"10.15622/SP.18.1.30-56","DOIUrl":"https://doi.org/10.15622/SP.18.1.30-56","url":null,"abstract":"Recently, Speech Emotion Recognition (SER) has become an important research topic of affective computing. It is a difficult problem, where some of the greatest challenges lie in the feature selection and representation tasks. A good feature representation should be able to reflect global trends as well as temporal structure of the signal, since emotions naturally evolve in time; it has become possible with the advent of Recurrent Neural Networks (RNN), which are actively used today for various sequence modeling tasks. This paper proposes a hybrid approach to feature representation, which combines traditionally engineered statistical features with Long Short-Term Memory (LSTM) sequence representation in order to take advantage of both short-term and long-term acoustic characteristics of the signal, therefore capturing not only the general trends but also temporal structure of the signal. The evaluation of the proposed method is done on three publicly available acted emotional speech corpora in three different languages, namely RUSLANA (Russian speech), BUEMODB (Turkish speech) and EMODB (German speech). Compared to the traditional approach, the results of our experiments show an absolute improvement of 2.3% and 2.8% for two out of three databases, and a comparative performance on the third. Therefore, provided enough training data, the proposed method proves effective in modelling emotional content of speech utterances.","PeriodicalId":53447,"journal":{"name":"SPIIRAS Proceedings","volume":"7 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2019-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"75167531","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The paper studies a problem of guaranteed transfer within a finite amount of time of a nonlinear dynamical system subjected to uncontrolled disturbances to a state where a given part of the variables equals zero. The bounded controls are offered to be generated by means of a feedback in form of nonlinear functions of phase variables of a given nonlinear controlled system of differential equations. The method of exact feedback linearization of the nonlinear system is used. As a result, the solution of the original nonlinear problem is narrowed down to solve the linear game-theoretic antagonistic control problem. Sufficient conditions are obtained with ensure that the problem has a guaranteed solution for the given domain of initial conditions. As an example, problem of the space turn of an asymmetric rigid bode (spacecraft) is considered within the framework of the method. Three reaction wheels are employed to produce necessary torque in the axes of the spacecraft. External uncontrolled disturbances, that have no statistical description, are taken into consideration in the process of reorientation. In this case the initial nonlinear controlled systems consists of dynamic Euler equations and Rodriges – Hamilton kinematic equations based on the quaternion parameterization of attitude kinematics. Two problems of the space turn of the spacecraft are considered. 1) The rest - to - rest reorientation problem. 2) The space turn from a stationary state to a given angular position; it is not assumed that the turn takes the spacecraft to a stationary state. The proposed approach allows common positions to give some already well-known solutions of these problems. A new solution of the reorientation problem is also given. For this new solution an estimation of the admissible domain of uncontrolled disturbances is found. Results of a numerical calculations are considered.
研究了非线性动力系统在有限时间内受不可控扰动的保证转移到给定部分变量为零的状态的问题。给出了用给定的非线性微分方程控制系统相变量非线性函数的反馈形式产生有界控制的方法。对非线性系统采用精确反馈线性化方法。将原非线性问题的求解范围缩小为求解线性博弈论对抗控制问题。得到了在给定初始条件域上保证问题有保证解的充分条件。以非对称刚体(航天器)的空间转向问题为例进行了分析。三个反作用轮被用来在航天器的轴上产生必要的扭矩。在重定向过程中考虑了没有统计描述的外部不受控制的干扰。在这种情况下,初始非线性控制系统由基于姿态运动学四元数参数化的动态欧拉方程和罗德里格斯-汉密尔顿运动学方程组成。考虑了航天器空间转向的两个问题。1) rest - to - rest重新定位问题。2)空间从静止状态转到给定角度位置;不假设转弯会使航天器进入静止状态。建议的方法允许共同立场给出这些问题的一些众所周知的解决方案。给出了一种新的重定向问题的解法。对于这个新解,给出了不受控制扰动容许域的估计。考虑了数值计算的结果。
{"title":"Feedback Liniarization Method for Problem of Control of a Part of Variables in Uncontrolled Disturbances","authors":"V. Vorotnikov, A. Vokhmyanina","doi":"10.15622/SP.61.3","DOIUrl":"https://doi.org/10.15622/SP.61.3","url":null,"abstract":"The paper studies a problem of guaranteed transfer within a finite amount of time of a nonlinear dynamical system subjected to uncontrolled disturbances to a state where a given part of the variables equals zero. The bounded controls are offered to be generated by means of a feedback in form of nonlinear functions of phase variables of a given nonlinear controlled system of differential equations. The method of exact feedback linearization of the nonlinear system is used. As a result, the solution of the original nonlinear problem is narrowed down to solve the linear game-theoretic antagonistic control problem. Sufficient conditions are obtained with ensure that the problem has a guaranteed solution for the given domain of initial conditions. \u0000As an example, problem of the space turn of an asymmetric rigid bode (spacecraft) is considered within the framework of the method. Three reaction wheels are employed to produce necessary torque in the axes of the spacecraft. External uncontrolled disturbances, that have no statistical description, are taken into consideration in the process of reorientation. In this case the initial nonlinear controlled systems consists of dynamic Euler equations and Rodriges – Hamilton kinematic equations based on the quaternion parameterization of attitude kinematics. Two problems of the space turn of the spacecraft are considered. 1) The rest - to - rest reorientation problem. 2) The space turn from a stationary state to a given angular position; it is not assumed that the turn takes the spacecraft to a stationary state. The proposed approach allows common positions to give some already well-known solutions of these problems. A new solution of the reorientation problem is also given. For this new solution an estimation of the admissible domain of uncontrolled disturbances is found. Results of a numerical calculations are considered.","PeriodicalId":53447,"journal":{"name":"SPIIRAS Proceedings","volume":"39 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2018-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"78972289","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
At present, the orbital constellations of satellites with the possibility of receiving, processing and relaying the ADS-B («Automatic Dependent Surveillance — Broadcast») system signals that ensure globality and continuity of the air traffic monitoring are being created and gradually enhanced. In accordance with the concept of ADS-B technology usage , each air traffic participant broadcasts its identity, location, and status parameters in broadcast mode. Due to the fact that the system was not designed to receive signals onboard the satellite, there are certain problems related to their energy availability, the presence of collisions of messages from different sources, the effect of Doppler effect and other factors. Developed simulation model of aviation surveillance systems based on the reception of signals containing identification and navigation information and transmitted over the air in the broadcasting mode. Software-implemented simulation algorithms allow to set various constraints and assumptions (radiation sources distribution, aviation communication systems signals receiving points, information transmitting channel, signals frequency and duration distribution models) and obtain the target indicators estimates of the space and ground-based aviation surveillance systems functioning, taking into account various spatial and energy factors and conditions for the radio signals propagation, and controlled objects actual placement and their movement dynamics in the world airspace. Simulation model use methods and examples for calculating the space and ground aviation surveillance systems functioning target indicators are presented.
{"title":"Simulation of Space and Ground-Based Aviation Surveillance Systems Functioning","authors":"Ya. A. Skorokhodov","doi":"10.15622/SP.61.2","DOIUrl":"https://doi.org/10.15622/SP.61.2","url":null,"abstract":"At present, the orbital constellations of satellites with the possibility of receiving, processing and relaying the ADS-B («Automatic Dependent Surveillance — Broadcast») system signals that ensure globality and continuity of the air traffic monitoring are being created and gradually enhanced. In accordance with the concept of ADS-B technology usage , each air traffic participant broadcasts its identity, location, and status parameters in broadcast mode. Due to the fact that the system was not designed to receive signals onboard the satellite, there are certain problems related to their energy availability, the presence of collisions of messages from different sources, the effect of Doppler effect and other factors. Developed simulation model of aviation surveillance systems based on the reception of signals containing identification and navigation information and transmitted over the air in the broadcasting mode. Software-implemented simulation algorithms allow to set various constraints and assumptions (radiation sources distribution, aviation communication systems signals receiving points, information transmitting channel, signals frequency and duration distribution models) and obtain the target indicators estimates of the space and ground-based aviation surveillance systems functioning, taking into account various spatial and energy factors and conditions for the radio signals propagation, and controlled objects actual placement and their movement dynamics in the world airspace. Simulation model use methods and examples for calculating the space and ground aviation surveillance systems functioning target indicators are presented.","PeriodicalId":53447,"journal":{"name":"SPIIRAS Proceedings","volume":"51 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2018-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"74536543","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Mathematical models of the Earth system and its components represent one of the most powerful and effective instruments applied to explore the Earth system's behaviour in the past and present, and to predict its future state considering external influence. These models are critically reliant on a large number of various observations (in situ and remotely sensed) since the prediction accuracy is determined by, amongst other things, the accuracy of the initial state of the system in question, which, in turn, is defined by observational data provided by many different instrument types. The development of an observing network is very costly, hence the estimation of the effectiveness of existing observation network and the design of a prospective one, is very important. The objectives of this paper are (1) to present the adjoint-based approach that allows us to estimate the impact of various observations on the accuracy of prediction of the Earth system and its components, and (2) to illustrate the application of this approach to two coupled low-order chaotic dynamical systems and to the ACCESS (Australian Community Climate and Earth System Simulator) global model used operationally in the Australian Bureau of Meteorology. The results of numerical experiments show that by using the adjoint-based method it is possible to rank the observations by the degree of their importance and also to estimate the influence of target observations on the quality of predictions.
{"title":"Observation Impact Assessment on the Prediction of the Earth System Dynamics Using the Adjoint-Based Method","authors":"P. Steinle, C. Tingwell, S. Soldatenko","doi":"10.15622/sp.61.1","DOIUrl":"https://doi.org/10.15622/sp.61.1","url":null,"abstract":"Mathematical models of the Earth system and its components represent one of the most powerful and effective instruments applied to explore the Earth system's behaviour in the past and present, and to predict its future state considering external influence. These models are critically reliant on a large number of various observations (in situ and remotely sensed) since the prediction accuracy is determined by, amongst other things, the accuracy of the initial state of the system in question, which, in turn, is defined by observational data provided by many different instrument types. The development of an observing network is very costly, hence the estimation of the effectiveness of existing observation network and the design of a prospective one, is very important. The objectives of this paper are (1) to present the adjoint-based approach that allows us to estimate the impact of various observations on the accuracy of prediction of the Earth system and its components, and (2) to illustrate the application of this approach to two coupled low-order chaotic dynamical systems and to the ACCESS (Australian Community Climate and Earth System Simulator) global model used operationally in the Australian Bureau of Meteorology. The results of numerical experiments show that by using the adjoint-based method it is possible to rank the observations by the degree of their importance and also to estimate the influence of target observations on the quality of predictions.","PeriodicalId":53447,"journal":{"name":"SPIIRAS Proceedings","volume":"1 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2018-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"74022692","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}