Pub Date : 2019-04-24DOI: 10.1109/SIU.2019.8806396
N. Yagmur, Baris Baykant Alagöz
Gradient descent dynamics is an optimization techniques that is widely used in machine learning applications. This technique updates model parameter in the direction of descending of learning error. In this study, Lyapunov stability of continuous time gradient descent dynamics is investigated and robust stability condition, which is needed for implementation of gradient descent dynamics in intelligent control system applications, is evaluated. In a illustrative example, for a De Jong's function type error function, solutions of continuous gradient descent dynamics and Euler method based numerical solutions are compared and stability concerns is discussed.
{"title":"Comparision of Solutions of Numerical Gradient Descent Method and Continous Time Gradient Descent Dynamics and Lyapunov Stability","authors":"N. Yagmur, Baris Baykant Alagöz","doi":"10.1109/SIU.2019.8806396","DOIUrl":"https://doi.org/10.1109/SIU.2019.8806396","url":null,"abstract":"Gradient descent dynamics is an optimization techniques that is widely used in machine learning applications. This technique updates model parameter in the direction of descending of learning error. In this study, Lyapunov stability of continuous time gradient descent dynamics is investigated and robust stability condition, which is needed for implementation of gradient descent dynamics in intelligent control system applications, is evaluated. In a illustrative example, for a De Jong's function type error function, solutions of continuous gradient descent dynamics and Euler method based numerical solutions are compared and stability concerns is discussed.","PeriodicalId":326275,"journal":{"name":"2019 27th Signal Processing and Communications Applications Conference (SIU)","volume":"6 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-04-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129865455","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 : 2019-04-24DOI: 10.1109/SIU.2019.8806348
A. Guney, Mustafa Namdar, Arif Basgumus
In this study, the performance of the angle of arrival (AoA) method, which is one of the location estimation algorithms, in 5G cellular networks is investigated. Sensitive location information is obtained with the help of the mathematical algorithms generated by taking advantage of the arrival angle of the signals emitted from the ultra-dense cells. In the proposed system model, the performance comparison of the least squares (LS) and maximum likelihood (ML) methods are given. It is found that the ML method has less position estimation error than the LS method, approximately 2 times in x axis and 3.5 times in y axis. The numerical results informed that the AoA location estimation algorithm can be used for a precise location information estimation in 5G cellular networks.
{"title":"Performance Comparison of LS and ML Methods for AoA Algorithm in 5G Cellular Networks","authors":"A. Guney, Mustafa Namdar, Arif Basgumus","doi":"10.1109/SIU.2019.8806348","DOIUrl":"https://doi.org/10.1109/SIU.2019.8806348","url":null,"abstract":"In this study, the performance of the angle of arrival (AoA) method, which is one of the location estimation algorithms, in 5G cellular networks is investigated. Sensitive location information is obtained with the help of the mathematical algorithms generated by taking advantage of the arrival angle of the signals emitted from the ultra-dense cells. In the proposed system model, the performance comparison of the least squares (LS) and maximum likelihood (ML) methods are given. It is found that the ML method has less position estimation error than the LS method, approximately 2 times in x axis and 3.5 times in y axis. The numerical results informed that the AoA location estimation algorithm can be used for a precise location information estimation in 5G cellular networks.","PeriodicalId":326275,"journal":{"name":"2019 27th Signal Processing and Communications Applications Conference (SIU)","volume":"63 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-04-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123832106","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 : 2019-04-24DOI: 10.1109/SIU.2019.8806550
H. Aydan, Çagatay Demirel, G. Ince, I. Koçak
Voice is still the main communication tool for us in our daily lives. Human posture is one of the important factors affecting the quality of sound and professional voice users have the ability to place their voices which is the ability to project ones voice to a location without necessarily turning there. An informatics based assistance system was developed to help voice therapists in their voice placement and posture improving therapies. This multi-modal support system which uses motion capture and multi array microphones to give feedback to doctors and patients in real time. In this work the efficiency of this system was demonstrated with user experience and usability tests done on human subjects.
{"title":"A Multi-Modal Support System for Voice Therapy","authors":"H. Aydan, Çagatay Demirel, G. Ince, I. Koçak","doi":"10.1109/SIU.2019.8806550","DOIUrl":"https://doi.org/10.1109/SIU.2019.8806550","url":null,"abstract":"Voice is still the main communication tool for us in our daily lives. Human posture is one of the important factors affecting the quality of sound and professional voice users have the ability to place their voices which is the ability to project ones voice to a location without necessarily turning there. An informatics based assistance system was developed to help voice therapists in their voice placement and posture improving therapies. This multi-modal support system which uses motion capture and multi array microphones to give feedback to doctors and patients in real time. In this work the efficiency of this system was demonstrated with user experience and usability tests done on human subjects.","PeriodicalId":326275,"journal":{"name":"2019 27th Signal Processing and Communications Applications Conference (SIU)","volume":"92 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-04-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123359844","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 : 2019-04-24DOI: 10.1109/SIU.2019.8806611
B. Demirel, Omer Özdil, Yunus Emre Esin, Şafak Öztürk
Hyperspectral images are obtained by dividing the electromagnetic spectrum into hundreds of narrow bands. Thanks to this feature, hyperspectral imaging is successful in distinguishing surface materials and is frequently used in target detection. In this study, long short-term memory and spectral angle mapper are used to detect targets in images obtained from the VNIR sensor. Deep neural networks require annotated data related to each target and also background classes for target detection in hyperspectral images. In this study, the background objects are eliminated by using the spectral angle mapper as a kind of filter, and the long short-term memory is only trained on the candidate target signatures. Therefore, data annotation activities are carried out only for candidate target classes and data annotation cost is reduced. In addition, the experimental results show that the long short-term memory model, which is trained on signatures collected from 30 meter heights, detects targets successfully independently of height.
{"title":"Hyperspectral Target Detection Using Long Short-Term Memory and Spectral Angle Mapper","authors":"B. Demirel, Omer Özdil, Yunus Emre Esin, Şafak Öztürk","doi":"10.1109/SIU.2019.8806611","DOIUrl":"https://doi.org/10.1109/SIU.2019.8806611","url":null,"abstract":"Hyperspectral images are obtained by dividing the electromagnetic spectrum into hundreds of narrow bands. Thanks to this feature, hyperspectral imaging is successful in distinguishing surface materials and is frequently used in target detection. In this study, long short-term memory and spectral angle mapper are used to detect targets in images obtained from the VNIR sensor. Deep neural networks require annotated data related to each target and also background classes for target detection in hyperspectral images. In this study, the background objects are eliminated by using the spectral angle mapper as a kind of filter, and the long short-term memory is only trained on the candidate target signatures. Therefore, data annotation activities are carried out only for candidate target classes and data annotation cost is reduced. In addition, the experimental results show that the long short-term memory model, which is trained on signatures collected from 30 meter heights, detects targets successfully independently of height.","PeriodicalId":326275,"journal":{"name":"2019 27th Signal Processing and Communications Applications Conference (SIU)","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-04-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121173544","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 : 2019-04-24DOI: 10.1109/SIU.2019.8806256
Eylem Erdogan, S. Çolak, Mustafa Namdar, Arif Basgumus, Hakan Alakoca, L. Durak-Ata
In this work, the adverse effects of a partial band jammer on a multi-user, multi-input multi-output cognitive radio networks are analyzed. In the proposed model, primary user transmits its information to the receiver with the aid of linear interference alignment method which uses a precoder and an interference suppression matrix. In this network, it is assumed that a partial band jammer adversely affects the transmission quality of the primary network. The analysis starts with the derivation of end-to-end signal to jammer noise ratio. Then, with the aid of probability density function and cumulative distribution functions, symbol error probability and outage probability are derived.
{"title":"Impact of Partial Band Jammer in Cognitive Radio Networks with Interference Alignment","authors":"Eylem Erdogan, S. Çolak, Mustafa Namdar, Arif Basgumus, Hakan Alakoca, L. Durak-Ata","doi":"10.1109/SIU.2019.8806256","DOIUrl":"https://doi.org/10.1109/SIU.2019.8806256","url":null,"abstract":"In this work, the adverse effects of a partial band jammer on a multi-user, multi-input multi-output cognitive radio networks are analyzed. In the proposed model, primary user transmits its information to the receiver with the aid of linear interference alignment method which uses a precoder and an interference suppression matrix. In this network, it is assumed that a partial band jammer adversely affects the transmission quality of the primary network. The analysis starts with the derivation of end-to-end signal to jammer noise ratio. Then, with the aid of probability density function and cumulative distribution functions, symbol error probability and outage probability are derived.","PeriodicalId":326275,"journal":{"name":"2019 27th Signal Processing and Communications Applications Conference (SIU)","volume":"50 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-04-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134361422","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 : 2019-04-24DOI: 10.1109/SIU.2019.8806456
Adil Kaan Akan, B. B. Kivilcim, Emre Akbas, Sharlene D. Newman, F. Yarman-Vural
It is hypothesized that the process of complex problem solving in human brain consists of two basic phases, namely, planning and execution. In this study, we propose a computational model in order to verify this hypothesis. For this purpose, we develop a holistic approach for decoding the planning and execution phases of complex problem solving, using the functional magnetic resonance imaging data (fMRI), recorded when the subjects play the Tower of London (TOL) game. In the first step of the proposed study, we estimate a brain network, called Artificial Brain Network (ABN), by designing an artificial neural network, whose weights correspond to the edge weights of the brain network established among the anatomic regions. Then, we decode the planning and execution tasks of complex problem slowing by training a multi-layer perceptron. It is shown that the edge weights of the artificial brain network capture the functional connectivity among anatomic brain regions. When trained on the edge weights of brain networks extracted from average BOLD activation of anatomical regions, the proposed model successfully discriminates the planning and execution phases of complex problem solving process. We compare the suggested computational brain network model to the state of the art models reported in the literature and observe that the decoding performance of the suggested model is better then the available methods in the literature.
{"title":"Modeling and Decoding Complex Problem Solving Process by Artificial Neural Networks","authors":"Adil Kaan Akan, B. B. Kivilcim, Emre Akbas, Sharlene D. Newman, F. Yarman-Vural","doi":"10.1109/SIU.2019.8806456","DOIUrl":"https://doi.org/10.1109/SIU.2019.8806456","url":null,"abstract":"It is hypothesized that the process of complex problem solving in human brain consists of two basic phases, namely, planning and execution. In this study, we propose a computational model in order to verify this hypothesis. For this purpose, we develop a holistic approach for decoding the planning and execution phases of complex problem solving, using the functional magnetic resonance imaging data (fMRI), recorded when the subjects play the Tower of London (TOL) game. In the first step of the proposed study, we estimate a brain network, called Artificial Brain Network (ABN), by designing an artificial neural network, whose weights correspond to the edge weights of the brain network established among the anatomic regions. Then, we decode the planning and execution tasks of complex problem slowing by training a multi-layer perceptron. It is shown that the edge weights of the artificial brain network capture the functional connectivity among anatomic brain regions. When trained on the edge weights of brain networks extracted from average BOLD activation of anatomical regions, the proposed model successfully discriminates the planning and execution phases of complex problem solving process. We compare the suggested computational brain network model to the state of the art models reported in the literature and observe that the decoding performance of the suggested model is better then the available methods in the literature.","PeriodicalId":326275,"journal":{"name":"2019 27th Signal Processing and Communications Applications Conference (SIU)","volume":"34 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-04-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121915182","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 : 2019-04-24DOI: 10.1109/SIU.2019.8806607
Uygar Demir, Baris Yuksekkaya, C. Toker
In this work, optimum transmit power allocation in V2X networks to maximize the aggregate data rate in the V2I network without violating individual peak transmit power constraints on the V2I users and the interference power constraints on the V2V users is considered. In the proposed model, V2I users form a multiple-access channel to the roadside unit and cause interference in the V2V network. Under this setup, it is first shown that the data rate maximizing optimum power allocation vector lies at one of the vertices of the feasible set of all transmit power vectors. The structure derived for the optimum power allocation vectors simplifies the solution of the power optimization problem significantly. That is, calculating and comparing the data rates at the vertices of the feasible power set, the optimum power allocation vector can be derived for each channel state. Furthermore, with the entry of each new V2V user to the system, the number of vertices increases at most by 3. Time division multiple access solution as a special case appears by a subset of our solution. In the final part of the paper, the theoretical results obtained are utilized to give numerical performance figures for V2X networks.
{"title":"Interference Aware Optimal Resource Allocation on V2X Networks","authors":"Uygar Demir, Baris Yuksekkaya, C. Toker","doi":"10.1109/SIU.2019.8806607","DOIUrl":"https://doi.org/10.1109/SIU.2019.8806607","url":null,"abstract":"In this work, optimum transmit power allocation in V2X networks to maximize the aggregate data rate in the V2I network without violating individual peak transmit power constraints on the V2I users and the interference power constraints on the V2V users is considered. In the proposed model, V2I users form a multiple-access channel to the roadside unit and cause interference in the V2V network. Under this setup, it is first shown that the data rate maximizing optimum power allocation vector lies at one of the vertices of the feasible set of all transmit power vectors. The structure derived for the optimum power allocation vectors simplifies the solution of the power optimization problem significantly. That is, calculating and comparing the data rates at the vertices of the feasible power set, the optimum power allocation vector can be derived for each channel state. Furthermore, with the entry of each new V2V user to the system, the number of vertices increases at most by 3. Time division multiple access solution as a special case appears by a subset of our solution. In the final part of the paper, the theoretical results obtained are utilized to give numerical performance figures for V2X networks.","PeriodicalId":326275,"journal":{"name":"2019 27th Signal Processing and Communications Applications Conference (SIU)","volume":"17 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-04-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124052616","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 : 2019-04-24DOI: 10.1109/SIU.2019.8806416
Ömer Kirbiyik, Enis Simsar, A. Cemgil
In this study, we compare deep learning methods for generating images of handwritten characters. This problem can be thought of as a restricted Turing test: A human draws a character from any desired alphabet and the system synthesizes images with similar appearances. The intention here is not to merely duplicate the input image but to add random perturbations to give the impression of being human-produced. For this purpose, the images produced by two different generative models (Generative Adversarial Network and Variational Autoencoder) and the related training method (Reptile) are examined with respect to their visual quality in a subjective manner. Also, the capability of transferring the knowledge that is obtained by the model is challenged by using different datasets for the training and test processes. Using the proposed model and meta-learning method, it is possible to produce not only images similar to the ones in the training set but also novel images that belong to a class which is seen for the first time.
{"title":"Comparison of Deep Generative Models for the Generation of Handwritten Character Images","authors":"Ömer Kirbiyik, Enis Simsar, A. Cemgil","doi":"10.1109/SIU.2019.8806416","DOIUrl":"https://doi.org/10.1109/SIU.2019.8806416","url":null,"abstract":"In this study, we compare deep learning methods for generating images of handwritten characters. This problem can be thought of as a restricted Turing test: A human draws a character from any desired alphabet and the system synthesizes images with similar appearances. The intention here is not to merely duplicate the input image but to add random perturbations to give the impression of being human-produced. For this purpose, the images produced by two different generative models (Generative Adversarial Network and Variational Autoencoder) and the related training method (Reptile) are examined with respect to their visual quality in a subjective manner. Also, the capability of transferring the knowledge that is obtained by the model is challenged by using different datasets for the training and test processes. Using the proposed model and meta-learning method, it is possible to produce not only images similar to the ones in the training set but also novel images that belong to a class which is seen for the first time.","PeriodicalId":326275,"journal":{"name":"2019 27th Signal Processing and Communications Applications Conference (SIU)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-04-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125797560","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 : 2019-04-24DOI: 10.1109/SIU.2019.8806518
Mehmet Can, I. Altunbas, E. Başar
In this paper, we aim to obtain better performance at high spectral efficiencies for downlink communication systems by combining media-based modulation (MBM) with nonorthogonal multiple access (NOMA). An union bound for the average bit error probability of the proposed system is derived in closed-form. The performance of the system has been investigated by using power allocation methods, which are constant and vary according to signal-to-noise ratio (SNR) values. It is shown that the proposed system provides better error performance compared to conventional NOMA systems, especially in high spectral efficiency. The accuracy of the theoretical analysis is verified by computer simulations.
{"title":"Media-Based Modulation Assisted Non-Orthogonal Multiple Access","authors":"Mehmet Can, I. Altunbas, E. Başar","doi":"10.1109/SIU.2019.8806518","DOIUrl":"https://doi.org/10.1109/SIU.2019.8806518","url":null,"abstract":"In this paper, we aim to obtain better performance at high spectral efficiencies for downlink communication systems by combining media-based modulation (MBM) with nonorthogonal multiple access (NOMA). An union bound for the average bit error probability of the proposed system is derived in closed-form. The performance of the system has been investigated by using power allocation methods, which are constant and vary according to signal-to-noise ratio (SNR) values. It is shown that the proposed system provides better error performance compared to conventional NOMA systems, especially in high spectral efficiency. The accuracy of the theoretical analysis is verified by computer simulations.","PeriodicalId":326275,"journal":{"name":"2019 27th Signal Processing and Communications Applications Conference (SIU)","volume":"169 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-04-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124721531","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 multi-agent system (MAS) is a network aimed to solve problems that exceed the individual capabilities of agents, without having a central control unit. Although wireless channels provide a more suitable application environment than the wired channels, the studies on the problems that may arise when MAS is implemented on wireless channels are limited in the literature. Wireless channel applications of MAS are more vulnerable to security problems or errors arising from the characteristics of the channels. This study aims to investigate the effect of jammers that can be encountered in wireless communication networks using Monte Carlo simulations under various performance criteria.
{"title":"The Effect of Jammers in Multi Agent Systems","authors":"Emre Durmaz, Gülay Öke Günel, G. Ascheid, Guido Dartmann, Günes Karabulut-Kurt","doi":"10.1109/SIU.2019.8806583","DOIUrl":"https://doi.org/10.1109/SIU.2019.8806583","url":null,"abstract":"A multi-agent system (MAS) is a network aimed to solve problems that exceed the individual capabilities of agents, without having a central control unit. Although wireless channels provide a more suitable application environment than the wired channels, the studies on the problems that may arise when MAS is implemented on wireless channels are limited in the literature. Wireless channel applications of MAS are more vulnerable to security problems or errors arising from the characteristics of the channels. This study aims to investigate the effect of jammers that can be encountered in wireless communication networks using Monte Carlo simulations under various performance criteria.","PeriodicalId":326275,"journal":{"name":"2019 27th Signal Processing and Communications Applications Conference (SIU)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-04-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130067883","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}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：