Impedance measurement has been widely used as an effective indicator for characterizing samples. ... more Impedance measurement has been widely used as an effective indicator for characterizing samples. Traditional high accuracy impedance analyzers are complicated, expensive, and non-portable. Many kinds of research require low cost, portable, and high precision impedance analyzer devices. AD5933 impedance analyzer integrated circuit has been popularly used for fulfilling these requirements. There are many successful applications of the AD5933 integrated circuit; however, the most significant drawback is nonlinear calibration requirements for high precision measurement in a specified range. The calibration and unknown impedances must be close enough to each other for better measurement accuracy. In the literature, calibration impedance arrays increasing the complexity and processing time are commonly used for high accuracy measurements. In this study, an artificial neural network-based signal post-processing algorithm is proposed to overcome the calibration requirements of the AD5933 integrated circuit, which requires different impedances for different ranges. In the literature, a neural network-based solution has not been applied to this phenomenon. An application specific artificial neural network topology is developed and trained for high precision impedance measurement using a fixed calibration impedance. The proposed measurement system is designed for operating in the range of nominal skin impedance. The average mean square error of measurements is obtained as 0.206%. Although a fixed calibration resistance is used, the proposed signal post-processing approach significantly improved the measurement accuracy of the AD5933 integrated circuit. The high accuracy measurement results prove the effectiveness of the proposed measurement system. The developed system offers portable, simple, and low cost high precision impedance analyzer.
2017 IEEE International Conference on INnovations in Intelligent SysTems and Applications (INISTA), 2017
Increase in popularity of deep convolutional neural networks in many different areas leads to inc... more Increase in popularity of deep convolutional neural networks in many different areas leads to increase in the use of these networks in reinforcement learning. Training a huge deep neural network structure by using simple gradient descent learning can take quite a long time. Some additional learning approaches should be utilized to solve this problem. One of these techniques is use of momentum which accelerates gradient descent learning. Although momentum techniques are mostly developed for supervised learning problems, it can also be used for reinforcement learning problems. However, its efficiency may vary due to the dissimilarities in two training learning processes. In this paper, the performances of different momentum techniques are compared for one of the reinforcement learning problems; Othello game benchmark. Test results show that the Nesterov accelerated momentum technique provided a more effective generalization on benchmark
Bu calismada, onceden tasarlanmis olan ozgun KNT/MOSFET tabanli aktif elektrotun ve yaygin olarak... more Bu calismada, onceden tasarlanmis olan ozgun KNT/MOSFET tabanli aktif elektrotun ve yaygin olarak kullanilan islak elektrotun elektrik devresi esdegerlerine, farkli biyopotansiyeller uygulanarak benzetimleri gerceklestirilmistir. Elektrotlarin giris sinyallerine verdigi yanitlar, belirli cilt yuzeyi kosullarinda karsilastirilmistir. Elektrotlari degerlendirmek icin elde edilen biyopotansiyellerin Fourier donusumleri ve toplam harmonik bozulmalari incelenmistir. Simulasyon sonuclari, KNT/MOSFET tabanli elektrotun islak elektrottan daha iyi sonuclar verdigini ve biyopotansiyelleri yuksek kalitede olcebilecegini gostermektedir.
Abstract The simulation of realistic device models in quantum transport requires an extreme amoun... more Abstract The simulation of realistic device models in quantum transport requires an extreme amount of memory and computation time. The computational burden in quantum transport is caused by the recursive numerical solution requirement of the Schrodinger equation with non-equilibrium Green’s function formalism. Ever decreasing device size increases the domination of the quantum mechanical effects such as scattering. Considerations of the quantum mechanical effects are crucial for emerging nanoscale devices. The solutions must consider the interactions between electron-electron, electron-phonon for qualified device modeling. In this work, a modified version of General Regression Neural Network and Non-Equilibrium Green’s Function hybrid modeling approach is proposed to overcome the mentioned computational burden. Through proposed computation processing, a pattern layer node is assigned for each atom in the atomic layer. In modified GRNN, pattern layer neuron values were extracted from NEGF calculation of three atomic layers. Higher atomic layer potential function calculation for Schrodinger equation is estimated by modified GRNN with dynamic pattern layer extension ability. A regression neuron is added to the output of the modified GRNN. Proposed modified GRNN topology is applied to model and solve atomic layer potential functions of Tunnel Field Effect Transistor in the range of seven to twenty-three atomic layers. Each atomic layer contains a hundred atoms in a row. Training data are obtained from the first three atomic layers of Tunnel FET. These training data are used for the estimation of test results for seven to twenty-three atomic layers. Results are compared with that of the incoherent NEGF model of Datta. Approximately 40% simulation convergence time decrease is observed during implementations. Simulation results proved the importance and efficiency of the proposed approach.
Journal of Information and Telecommunication, 2018
Recently, the popularity of deep artificial neural networks has increased considerably. Generally... more Recently, the popularity of deep artificial neural networks has increased considerably. Generally, the method used in the training of these structures is simple gradient descent. However, training a deep structure with simple gradient descent can take quite a long time. Some additional approaches have been utilized to solve this problem. One of these techniques is the momentum that accelerates gradient descent learning. Momentum techniques can be used for supervised learning as well as for reinforcement learning. However, its efficiency may vary due to the dissimilarities in two learning processes. While the expected values of inputs are clearly known in supervised learning, it may take long-running iterations to reach the exact expected values of the states in reinforcement learning. In an online learning approach, a deep neural network should not memorize and continue to converge with the more precise values that exist over time during these iterations. For this reason, it is necessary to use a momentum technique that both adapt to reinforcement learning and accelerate the learning process. In this paper, the performance of different momentum techniques is compared with the Othello game benchmark. Test results show that the Nesterov momentum technique provided a more effective generalization with an online reinforcement learning approach.
Density Functional Theory (DFT) calculations used in the Carbon Nanotubes (CNT) design take a ver... more Density Functional Theory (DFT) calculations used in the Carbon Nanotubes (CNT) design take a very long time even in the simulation environment as it is well known in literature. In this study, calculation time of DFT for geometry optimization of CNT is reduced from days to minutes using seven artificial intelligence-based and one statistical-based methods and the results are compared. The best results are achieved from ANFIS and ANN based models and these models can be used instead of CNT simulation software with high accuracy.
ABSTRACT Although sunshine duration (SD) is one of the most frequently measured meteorological pa... more ABSTRACT Although sunshine duration (SD) is one of the most frequently measured meteorological parameters, there is a lack of measurements in some parts of the world. Hence, it should be estimated accurately for areas where no reliable measurement is possible. The main objective of this study is to evaluate the potential of support vector machine (SVM) approach for estimating daily SD. For this purpose, three different kernels of SVM, such as linear, polynomial, and radial basis function (RBF), were used. Different combinations of five related meteorological parameters, namely cloud cover, maximum temperature (Tmax), minimum temperature (Tmin), relative humidity (RH), and wind speed (WS), and one astronomic parameter, day length, were considered as the inputs of the models, and the output was obtained as daily SD. Simulated values of the models were compared with ground measured values, and concluded that the usage of the SVM-RBF estimator with combination of all input attributes produced the best results. The coefficient of determination, root mean square error, and mean absolute error were found to be 0.8435, 1.5105 h, and 1.0771 h, respectively, for the pooled four-year daily data set of 14 stations in Turkey. It was also deduced that accuracy increased as the number of attributes increased and the major contribution to this came from RH as compared with Tmax, Tmin, and WS. This study has shown that the SVM methodology can be a good alternative for conventional and artificial neural network methods for estimating daily SD.
In this paper, four artificial neural network (ANN) models [i.e., feed-forward neural network (FF... more In this paper, four artificial neural network (ANN) models [i.e., feed-forward neural network (FFNN), function fitting neural network (FITNET), cascade-forward neural network (CFNN) and generalized regression neural network] have been developed for atomic coordinate prediction of carbon nanotubes (CNTs). The research reported in this study has two primary objectives: (1) to develop ANN prediction models that calculate atomic coordinates of CNTs instead of using any simulation software and (2) to use results of the ANN models as an initial value of atomic coordinates for reducing number of iterations in calculation process. The dataset consisting of 10,721 data samples was created by combining the atomic coordinates of elements and chiral vectors using BIOVIA Materials Studio CASTEP (CASTEP) software. All prediction models yield very low mean squared normalized error and mean absolute error rates. Multiple correlation coefficient (R) results of FITNET, FFNN and CFNN models are close to 1. Compared with CASTEP, calculation times decrease from days to minutes. It would seem possible to predict CNTs’ atomic coordinates using ANN models can be successfully used instead of mathematical calculations.
Memory fragmentation is a serious obstacle preventing efficient memory usage. Garbage collectors ... more Memory fragmentation is a serious obstacle preventing efficient memory usage. Garbage collectors may solve the problem; however, they cause serious performance impact, memory and energy consumption. Therefore, various memory allocators have been developed. Software developers must test memory allocators, and find an efficient one for their programs. Instead of this cumbersome method, we propose a novel approach for dynamically deciding the best memory allocator for every application. The proposed solution tests each process with various memory allocators. After the testing, it selects an efficient memory allocator according to condition of operating system (OS). If OS runs out of memory, then it selects the most memory efficient allocator for new processes. If most of the CPU power was occupied, then it selects the fastest allocator. Otherwise, the balanced allocator is selected. According to test results, the proposed solution offers up to 58% less fragmented memory, and 90% faster memory operations. In average of 107 processes, it offers 7.16?2.53% less fragmented memory, and 1.79?7.32% faster memory operations. The test results also prove the proposed approach is unbeatable by any memory allocator. In conclusion, the proposed method is a dynamic and efficient solution to the memory fragmentation problem. HighlightsOur solution is an intelligent memory allocator selector for operating systems.The solution selects an efficient and fastest memory allocator for each process.The approach reduces memory fragmentation, and increases system performance.Our solution is a dynamic and efficient solution to memory fragmentation problem.
In this work, parametric deviation of a process with double poly is investigated. A single Floati... more In this work, parametric deviation of a process with double poly is investigated. A single Floating Gate MOS (FGMOS) transistor is taken into consideration for simulating the process error. The parametric deviation effects on the n-channel FGMOS drain current are visualized with the SPICE simulations for AMIS 0.5µ process technology.
In this work a new radial basis function based classification neural network named as generalized... more In this work a new radial basis function based classification neural network named as generalized classifier neural network, is proposed. The proposed generalized classifier neural network has five layers, unlike other radial basis function based neural networks such as generalized regression neural network and probabilistic neural network. They are input, pattern, summation, normalization and output layers. In addition to topological difference, the proposed neural network has gradient descent based optimization of smoothing parameter approach and diverge effect term added calculation improvements. Diverge effect term is an improvement on summation layer calculation to supply additional separation ability and flexibility. Performance of generalized classifier neural network is compared with that of the probabilistic neural network, multilayer perceptron algorithm and radial basis function neural network on 9 different data sets and with that of generalized regression neural network on 3 different data sets include only two classes in MATLAB environment. Better classification performance up to %89 is observed. Improved classification performances proved the effectivity of the proposed neural network.
Machine learning is a knowledge area starting at the point when data are explained or estimations... more Machine learning is a knowledge area starting at the point when data are explained or estimations are produced for the future. It generates functional approximation or classification models for the data. Various methods and algorithms form the base of machine learning. Everyday ...
Technological improvements lead big data producing, processing and storing systems. These systems... more Technological improvements lead big data producing, processing and storing systems. These systems must contain extraordinary capabilities to overcome complexity of the big data. Therefore, the methodologies utilized for data analysis have been evolved due to the increase in importance of extracting information from big data. Healthcare systems are important systems dealing with big data analysis. Deep learning is the most applied data analysis method. It becomes one of the most popular and up-to-date artificial neural network types with deep representation ability. Another powerful ability of deep learning is providing feature learning through convolutional neural networks. Deep learning has wide implementation areas in medical applications from diagnosis to treatment. Various deep learning methods are applied to the biomedical problems. In many applications, deep learning solutions are modified in accordance with the requirements of the problems. Through this chapter the most popul...
This paper introduces an artificial neural network (ANN) approach for estimating monthly mean dai... more This paper introduces an artificial neural network (ANN) approach for estimating monthly mean daily values of global sunshine duration (SD) for Turkey. Three different ANN models, namely, GRNN, MLP, and RBF, were used in the estimation processes. A climatic variable (cloud cover) and two geographical variables (day length and month) were used as input parameters in order to obtain monthly mean SD as output. The datasets of 34 stations which spread across Turkey were split into two parts. First part covering 21 years (1980-2000) was used for training and second part covering last six years (2001-2006) was used for testing. Statistical indicators have shown that, GRNN and MLP models produced better results than the RBF model and can be used safely for the estimation of monthly mean SD.
In this work, development of a voltage dependent resistance model for metallic carbon nanotubes i... more In this work, development of a voltage dependent resistance model for metallic carbon nanotubes is aimed. Firstly, the resistance of metallic carbon nanotube interconnects are obtained from ab initio simulations and then the voltage dependence of the resistance is modeled through regression. Selfconsistent non-equilibrium Green's function formalism combined with density functional theory is used for calculating the voltage dependent resistance of metallic carbon nanotubes. It is shown that voltage dependent resistances of carbon nanotubes can be accurately modeled as a polynomial function which enables rapid integration of carbon nanotube interconnect models into electronic design automation tools.
Çukurova Üniversitesi Mühendislik-Mimarlık Fakültesi Dergisi
Bu çalışmada, yeraltında üretim, nakliye ve depolama işleri sırasında biriken kömür tozlarının pa... more Bu çalışmada, yeraltında üretim, nakliye ve depolama işleri sırasında biriken kömür tozlarının patlama sınırlarını belirleyerek patlamanın önlenmesi amacıyla serpilecek taş tozu miktarının belirlenmesi için geliştirilen kamera görüntüsü tabanlı sistemle yapılan deneylerden elde edilen sonuçlarla laboratuvarda elde edilen taş tozu-kömür tozu karışımlarının sonuçları karşılaştırılmıştır. Deneylerde Zonguldak/Kozlu Bölgesinden temin edilen kömür tozları kullanılmıştır. Kamera görüntüsü Tabanlı Ölçme Sistemi, deney numunelerindeki taş tozu oranlarını en fazla %0,026 hata ile tanımlamaktadır.
Standard electrodes for electrophysiological signal acquisition in clinical applications such as ... more Standard electrodes for electrophysiological signal acquisition in clinical applications such as electrocardiography or electromyography require the use of electrolytic gel and skin abrasion for better electrical interface between electrode and skin. Since gel dries out over a period of time, signal deterioration takes place on long duration with wet electrodes. Dry electrodes are promising alternative for long duration recording; nevertheless, they suffer from high contact impedance and motion artifacts. The proposed work introduces a novel multi-walled carbon nanotube (MWCNT) modified metal oxide semiconductor field effect transistor (MOSFET) based electrode for electrophysiological measurements on human skin. Vertically aligned metallic MWCNTs grown on the gate of MOSFETs form the contact surface of the electrode. MWCNTs penetrate the outer layer of skin for stable and improved electrical contact without the gel. The proposed electrode utilizes advantages of the MOSFET such as direct charge– current conversion, insulation between skin and instrumentation unit, and low noise pre-amplification of electrophysiological signals. Electrical equivalent of MWCNTs, design, and microfabrication of convenient MOSFET are reported. MOSFET parameters are obtained from technology computer aided design simulation environment and combined with MWCNT parameters in the simulation program for integrated circuits emphasis. Simulated results of the proposed electrode exhibited lower contact impedance and high quality signal capture with respect to wet electrodes. The results show that the proposed electrode can be used for long duration recording of biopotentials with very high stable performance.
Impedance measurement has been widely used as an effective indicator for characterizing samples. ... more Impedance measurement has been widely used as an effective indicator for characterizing samples. Traditional high accuracy impedance analyzers are complicated, expensive, and non-portable. Many kinds of research require low cost, portable, and high precision impedance analyzer devices. AD5933 impedance analyzer integrated circuit has been popularly used for fulfilling these requirements. There are many successful applications of the AD5933 integrated circuit; however, the most significant drawback is nonlinear calibration requirements for high precision measurement in a specified range. The calibration and unknown impedances must be close enough to each other for better measurement accuracy. In the literature, calibration impedance arrays increasing the complexity and processing time are commonly used for high accuracy measurements. In this study, an artificial neural network-based signal post-processing algorithm is proposed to overcome the calibration requirements of the AD5933 integrated circuit, which requires different impedances for different ranges. In the literature, a neural network-based solution has not been applied to this phenomenon. An application specific artificial neural network topology is developed and trained for high precision impedance measurement using a fixed calibration impedance. The proposed measurement system is designed for operating in the range of nominal skin impedance. The average mean square error of measurements is obtained as 0.206%. Although a fixed calibration resistance is used, the proposed signal post-processing approach significantly improved the measurement accuracy of the AD5933 integrated circuit. The high accuracy measurement results prove the effectiveness of the proposed measurement system. The developed system offers portable, simple, and low cost high precision impedance analyzer.
2017 IEEE International Conference on INnovations in Intelligent SysTems and Applications (INISTA), 2017
Increase in popularity of deep convolutional neural networks in many different areas leads to inc... more Increase in popularity of deep convolutional neural networks in many different areas leads to increase in the use of these networks in reinforcement learning. Training a huge deep neural network structure by using simple gradient descent learning can take quite a long time. Some additional learning approaches should be utilized to solve this problem. One of these techniques is use of momentum which accelerates gradient descent learning. Although momentum techniques are mostly developed for supervised learning problems, it can also be used for reinforcement learning problems. However, its efficiency may vary due to the dissimilarities in two training learning processes. In this paper, the performances of different momentum techniques are compared for one of the reinforcement learning problems; Othello game benchmark. Test results show that the Nesterov accelerated momentum technique provided a more effective generalization on benchmark
Bu calismada, onceden tasarlanmis olan ozgun KNT/MOSFET tabanli aktif elektrotun ve yaygin olarak... more Bu calismada, onceden tasarlanmis olan ozgun KNT/MOSFET tabanli aktif elektrotun ve yaygin olarak kullanilan islak elektrotun elektrik devresi esdegerlerine, farkli biyopotansiyeller uygulanarak benzetimleri gerceklestirilmistir. Elektrotlarin giris sinyallerine verdigi yanitlar, belirli cilt yuzeyi kosullarinda karsilastirilmistir. Elektrotlari degerlendirmek icin elde edilen biyopotansiyellerin Fourier donusumleri ve toplam harmonik bozulmalari incelenmistir. Simulasyon sonuclari, KNT/MOSFET tabanli elektrotun islak elektrottan daha iyi sonuclar verdigini ve biyopotansiyelleri yuksek kalitede olcebilecegini gostermektedir.
Abstract The simulation of realistic device models in quantum transport requires an extreme amoun... more Abstract The simulation of realistic device models in quantum transport requires an extreme amount of memory and computation time. The computational burden in quantum transport is caused by the recursive numerical solution requirement of the Schrodinger equation with non-equilibrium Green’s function formalism. Ever decreasing device size increases the domination of the quantum mechanical effects such as scattering. Considerations of the quantum mechanical effects are crucial for emerging nanoscale devices. The solutions must consider the interactions between electron-electron, electron-phonon for qualified device modeling. In this work, a modified version of General Regression Neural Network and Non-Equilibrium Green’s Function hybrid modeling approach is proposed to overcome the mentioned computational burden. Through proposed computation processing, a pattern layer node is assigned for each atom in the atomic layer. In modified GRNN, pattern layer neuron values were extracted from NEGF calculation of three atomic layers. Higher atomic layer potential function calculation for Schrodinger equation is estimated by modified GRNN with dynamic pattern layer extension ability. A regression neuron is added to the output of the modified GRNN. Proposed modified GRNN topology is applied to model and solve atomic layer potential functions of Tunnel Field Effect Transistor in the range of seven to twenty-three atomic layers. Each atomic layer contains a hundred atoms in a row. Training data are obtained from the first three atomic layers of Tunnel FET. These training data are used for the estimation of test results for seven to twenty-three atomic layers. Results are compared with that of the incoherent NEGF model of Datta. Approximately 40% simulation convergence time decrease is observed during implementations. Simulation results proved the importance and efficiency of the proposed approach.
Journal of Information and Telecommunication, 2018
Recently, the popularity of deep artificial neural networks has increased considerably. Generally... more Recently, the popularity of deep artificial neural networks has increased considerably. Generally, the method used in the training of these structures is simple gradient descent. However, training a deep structure with simple gradient descent can take quite a long time. Some additional approaches have been utilized to solve this problem. One of these techniques is the momentum that accelerates gradient descent learning. Momentum techniques can be used for supervised learning as well as for reinforcement learning. However, its efficiency may vary due to the dissimilarities in two learning processes. While the expected values of inputs are clearly known in supervised learning, it may take long-running iterations to reach the exact expected values of the states in reinforcement learning. In an online learning approach, a deep neural network should not memorize and continue to converge with the more precise values that exist over time during these iterations. For this reason, it is necessary to use a momentum technique that both adapt to reinforcement learning and accelerate the learning process. In this paper, the performance of different momentum techniques is compared with the Othello game benchmark. Test results show that the Nesterov momentum technique provided a more effective generalization with an online reinforcement learning approach.
Density Functional Theory (DFT) calculations used in the Carbon Nanotubes (CNT) design take a ver... more Density Functional Theory (DFT) calculations used in the Carbon Nanotubes (CNT) design take a very long time even in the simulation environment as it is well known in literature. In this study, calculation time of DFT for geometry optimization of CNT is reduced from days to minutes using seven artificial intelligence-based and one statistical-based methods and the results are compared. The best results are achieved from ANFIS and ANN based models and these models can be used instead of CNT simulation software with high accuracy.
ABSTRACT Although sunshine duration (SD) is one of the most frequently measured meteorological pa... more ABSTRACT Although sunshine duration (SD) is one of the most frequently measured meteorological parameters, there is a lack of measurements in some parts of the world. Hence, it should be estimated accurately for areas where no reliable measurement is possible. The main objective of this study is to evaluate the potential of support vector machine (SVM) approach for estimating daily SD. For this purpose, three different kernels of SVM, such as linear, polynomial, and radial basis function (RBF), were used. Different combinations of five related meteorological parameters, namely cloud cover, maximum temperature (Tmax), minimum temperature (Tmin), relative humidity (RH), and wind speed (WS), and one astronomic parameter, day length, were considered as the inputs of the models, and the output was obtained as daily SD. Simulated values of the models were compared with ground measured values, and concluded that the usage of the SVM-RBF estimator with combination of all input attributes produced the best results. The coefficient of determination, root mean square error, and mean absolute error were found to be 0.8435, 1.5105 h, and 1.0771 h, respectively, for the pooled four-year daily data set of 14 stations in Turkey. It was also deduced that accuracy increased as the number of attributes increased and the major contribution to this came from RH as compared with Tmax, Tmin, and WS. This study has shown that the SVM methodology can be a good alternative for conventional and artificial neural network methods for estimating daily SD.
In this paper, four artificial neural network (ANN) models [i.e., feed-forward neural network (FF... more In this paper, four artificial neural network (ANN) models [i.e., feed-forward neural network (FFNN), function fitting neural network (FITNET), cascade-forward neural network (CFNN) and generalized regression neural network] have been developed for atomic coordinate prediction of carbon nanotubes (CNTs). The research reported in this study has two primary objectives: (1) to develop ANN prediction models that calculate atomic coordinates of CNTs instead of using any simulation software and (2) to use results of the ANN models as an initial value of atomic coordinates for reducing number of iterations in calculation process. The dataset consisting of 10,721 data samples was created by combining the atomic coordinates of elements and chiral vectors using BIOVIA Materials Studio CASTEP (CASTEP) software. All prediction models yield very low mean squared normalized error and mean absolute error rates. Multiple correlation coefficient (R) results of FITNET, FFNN and CFNN models are close to 1. Compared with CASTEP, calculation times decrease from days to minutes. It would seem possible to predict CNTs’ atomic coordinates using ANN models can be successfully used instead of mathematical calculations.
Memory fragmentation is a serious obstacle preventing efficient memory usage. Garbage collectors ... more Memory fragmentation is a serious obstacle preventing efficient memory usage. Garbage collectors may solve the problem; however, they cause serious performance impact, memory and energy consumption. Therefore, various memory allocators have been developed. Software developers must test memory allocators, and find an efficient one for their programs. Instead of this cumbersome method, we propose a novel approach for dynamically deciding the best memory allocator for every application. The proposed solution tests each process with various memory allocators. After the testing, it selects an efficient memory allocator according to condition of operating system (OS). If OS runs out of memory, then it selects the most memory efficient allocator for new processes. If most of the CPU power was occupied, then it selects the fastest allocator. Otherwise, the balanced allocator is selected. According to test results, the proposed solution offers up to 58% less fragmented memory, and 90% faster memory operations. In average of 107 processes, it offers 7.16?2.53% less fragmented memory, and 1.79?7.32% faster memory operations. The test results also prove the proposed approach is unbeatable by any memory allocator. In conclusion, the proposed method is a dynamic and efficient solution to the memory fragmentation problem. HighlightsOur solution is an intelligent memory allocator selector for operating systems.The solution selects an efficient and fastest memory allocator for each process.The approach reduces memory fragmentation, and increases system performance.Our solution is a dynamic and efficient solution to memory fragmentation problem.
In this work, parametric deviation of a process with double poly is investigated. A single Floati... more In this work, parametric deviation of a process with double poly is investigated. A single Floating Gate MOS (FGMOS) transistor is taken into consideration for simulating the process error. The parametric deviation effects on the n-channel FGMOS drain current are visualized with the SPICE simulations for AMIS 0.5µ process technology.
In this work a new radial basis function based classification neural network named as generalized... more In this work a new radial basis function based classification neural network named as generalized classifier neural network, is proposed. The proposed generalized classifier neural network has five layers, unlike other radial basis function based neural networks such as generalized regression neural network and probabilistic neural network. They are input, pattern, summation, normalization and output layers. In addition to topological difference, the proposed neural network has gradient descent based optimization of smoothing parameter approach and diverge effect term added calculation improvements. Diverge effect term is an improvement on summation layer calculation to supply additional separation ability and flexibility. Performance of generalized classifier neural network is compared with that of the probabilistic neural network, multilayer perceptron algorithm and radial basis function neural network on 9 different data sets and with that of generalized regression neural network on 3 different data sets include only two classes in MATLAB environment. Better classification performance up to %89 is observed. Improved classification performances proved the effectivity of the proposed neural network.
Machine learning is a knowledge area starting at the point when data are explained or estimations... more Machine learning is a knowledge area starting at the point when data are explained or estimations are produced for the future. It generates functional approximation or classification models for the data. Various methods and algorithms form the base of machine learning. Everyday ...
Technological improvements lead big data producing, processing and storing systems. These systems... more Technological improvements lead big data producing, processing and storing systems. These systems must contain extraordinary capabilities to overcome complexity of the big data. Therefore, the methodologies utilized for data analysis have been evolved due to the increase in importance of extracting information from big data. Healthcare systems are important systems dealing with big data analysis. Deep learning is the most applied data analysis method. It becomes one of the most popular and up-to-date artificial neural network types with deep representation ability. Another powerful ability of deep learning is providing feature learning through convolutional neural networks. Deep learning has wide implementation areas in medical applications from diagnosis to treatment. Various deep learning methods are applied to the biomedical problems. In many applications, deep learning solutions are modified in accordance with the requirements of the problems. Through this chapter the most popul...
This paper introduces an artificial neural network (ANN) approach for estimating monthly mean dai... more This paper introduces an artificial neural network (ANN) approach for estimating monthly mean daily values of global sunshine duration (SD) for Turkey. Three different ANN models, namely, GRNN, MLP, and RBF, were used in the estimation processes. A climatic variable (cloud cover) and two geographical variables (day length and month) were used as input parameters in order to obtain monthly mean SD as output. The datasets of 34 stations which spread across Turkey were split into two parts. First part covering 21 years (1980-2000) was used for training and second part covering last six years (2001-2006) was used for testing. Statistical indicators have shown that, GRNN and MLP models produced better results than the RBF model and can be used safely for the estimation of monthly mean SD.
In this work, development of a voltage dependent resistance model for metallic carbon nanotubes i... more In this work, development of a voltage dependent resistance model for metallic carbon nanotubes is aimed. Firstly, the resistance of metallic carbon nanotube interconnects are obtained from ab initio simulations and then the voltage dependence of the resistance is modeled through regression. Selfconsistent non-equilibrium Green's function formalism combined with density functional theory is used for calculating the voltage dependent resistance of metallic carbon nanotubes. It is shown that voltage dependent resistances of carbon nanotubes can be accurately modeled as a polynomial function which enables rapid integration of carbon nanotube interconnect models into electronic design automation tools.
Çukurova Üniversitesi Mühendislik-Mimarlık Fakültesi Dergisi
Bu çalışmada, yeraltında üretim, nakliye ve depolama işleri sırasında biriken kömür tozlarının pa... more Bu çalışmada, yeraltında üretim, nakliye ve depolama işleri sırasında biriken kömür tozlarının patlama sınırlarını belirleyerek patlamanın önlenmesi amacıyla serpilecek taş tozu miktarının belirlenmesi için geliştirilen kamera görüntüsü tabanlı sistemle yapılan deneylerden elde edilen sonuçlarla laboratuvarda elde edilen taş tozu-kömür tozu karışımlarının sonuçları karşılaştırılmıştır. Deneylerde Zonguldak/Kozlu Bölgesinden temin edilen kömür tozları kullanılmıştır. Kamera görüntüsü Tabanlı Ölçme Sistemi, deney numunelerindeki taş tozu oranlarını en fazla %0,026 hata ile tanımlamaktadır.
Standard electrodes for electrophysiological signal acquisition in clinical applications such as ... more Standard electrodes for electrophysiological signal acquisition in clinical applications such as electrocardiography or electromyography require the use of electrolytic gel and skin abrasion for better electrical interface between electrode and skin. Since gel dries out over a period of time, signal deterioration takes place on long duration with wet electrodes. Dry electrodes are promising alternative for long duration recording; nevertheless, they suffer from high contact impedance and motion artifacts. The proposed work introduces a novel multi-walled carbon nanotube (MWCNT) modified metal oxide semiconductor field effect transistor (MOSFET) based electrode for electrophysiological measurements on human skin. Vertically aligned metallic MWCNTs grown on the gate of MOSFETs form the contact surface of the electrode. MWCNTs penetrate the outer layer of skin for stable and improved electrical contact without the gel. The proposed electrode utilizes advantages of the MOSFET such as direct charge– current conversion, insulation between skin and instrumentation unit, and low noise pre-amplification of electrophysiological signals. Electrical equivalent of MWCNTs, design, and microfabrication of convenient MOSFET are reported. MOSFET parameters are obtained from technology computer aided design simulation environment and combined with MWCNT parameters in the simulation program for integrated circuits emphasis. Simulated results of the proposed electrode exhibited lower contact impedance and high quality signal capture with respect to wet electrodes. The results show that the proposed electrode can be used for long duration recording of biopotentials with very high stable performance.
Uploads
Papers by Mutlu Avci