Pub Date : 2020-12-20DOI: 10.1109/STA50679.2020.9329305
J. Snoussi, S. B. Elghali, M. Mimouni
The estimation of batteries State of charge is a crucial step in the developing of advanced plug-in and hybrid electric vehicles. In fact, the the accuracy of on line SOC estimation techniques is closely related to the reliability of the battery model which could efficiently describe the complex behavior of the battery during vehicle operation and rest periods. In this context, a new battery model is proposed and an online identification technique is developed to truck the model parameters variations and to ensure a high level of accuracy for onboard SOC estimation tasks. The accuracy of the developed model is verified by simulations using Matlab software and by experiments tests using a National Instruments platform.
{"title":"New LiFePO4 Battery Model Identification for Online SOC Estimation Application","authors":"J. Snoussi, S. B. Elghali, M. Mimouni","doi":"10.1109/STA50679.2020.9329305","DOIUrl":"https://doi.org/10.1109/STA50679.2020.9329305","url":null,"abstract":"The estimation of batteries State of charge is a crucial step in the developing of advanced plug-in and hybrid electric vehicles. In fact, the the accuracy of on line SOC estimation techniques is closely related to the reliability of the battery model which could efficiently describe the complex behavior of the battery during vehicle operation and rest periods. In this context, a new battery model is proposed and an online identification technique is developed to truck the model parameters variations and to ensure a high level of accuracy for onboard SOC estimation tasks. The accuracy of the developed model is verified by simulations using Matlab software and by experiments tests using a National Instruments platform.","PeriodicalId":158545,"journal":{"name":"2020 20th International Conference on Sciences and Techniques of Automatic Control and Computer Engineering (STA)","volume":"46 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-12-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126662475","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 : 2020-12-20DOI: 10.1109/STA50679.2020.9329355
Nessrine Abbassi, Rabie Helaly, Mohamed Ali Hajjaji, A. Mtibaa
after studying the pretrained VGGNet 19 model, we figured out that this model contains a large number of parameters that tend likely towards overfitting, which blocks the face expression recognition performance. This indicates that there is always some room for improvement. In this manuscript, we propose a new approach based on the VGGNet-19 network, in which we use several convolution layers with small filters and a dropout strategy. In the adopted model, the addition of convolution layers is recommended in order to give more precision to image classification. The experiment results suggest that the proposed model give promising results.
{"title":"A Deep Learning Facial Emotion Classification system: a VGGNet-19 based approach","authors":"Nessrine Abbassi, Rabie Helaly, Mohamed Ali Hajjaji, A. Mtibaa","doi":"10.1109/STA50679.2020.9329355","DOIUrl":"https://doi.org/10.1109/STA50679.2020.9329355","url":null,"abstract":"after studying the pretrained VGGNet 19 model, we figured out that this model contains a large number of parameters that tend likely towards overfitting, which blocks the face expression recognition performance. This indicates that there is always some room for improvement. In this manuscript, we propose a new approach based on the VGGNet-19 network, in which we use several convolution layers with small filters and a dropout strategy. In the adopted model, the addition of convolution layers is recommended in order to give more precision to image classification. The experiment results suggest that the proposed model give promising results.","PeriodicalId":158545,"journal":{"name":"2020 20th International Conference on Sciences and Techniques of Automatic Control and Computer Engineering (STA)","volume":"106 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-12-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116721306","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 : 2020-12-20DOI: 10.1109/STA50679.2020.9329302
Rabie Helaly, Mohamed Ali Hajjaji, F. M'sahli, A. Mtibaa
In this paper, we present a facial recognition system based on deep learning model. The proposed work is implemented on the embedded system named Raspberry Pi 4. For this, the “Xception convolutional neural network” model is chosen to achieve our emotion system recognition. The registered facial images are taken as input into the classifiers in the embedded system which classifies them into seven facial expressions. For the conduction of the experiment “Fer 2013” data set is used. The proposed model gives an accuracy of 94 % in Graphics Processing Unit” (GPU). After his implementation on the embedded system, according to his limitation against GPU performances, The accuracy is 89% on Raspberry Pi 4. Comparing to other recent works
本文提出了一种基于深度学习模型的人脸识别系统。所提出的工作在嵌入式系统Raspberry Pi 4上实现。为此，我们选择了“异常卷积神经网络”模型来实现我们的情感系统识别。将注册好的人脸图像输入到嵌入式系统的分类器中，分类器将其分类为7种面部表情。为了进行实验，使用了“Fer 2013”数据集。该模型在图形处理单元(GPU)上的准确率达到94%。在嵌入式系统上实现后，根据他对GPU性能的限制，在Raspberry Pi 4上的准确率为89%。与其他近期作品相比
{"title":"Deep Convolution Neural Network Implementation for Emotion Recognition System","authors":"Rabie Helaly, Mohamed Ali Hajjaji, F. M'sahli, A. Mtibaa","doi":"10.1109/STA50679.2020.9329302","DOIUrl":"https://doi.org/10.1109/STA50679.2020.9329302","url":null,"abstract":"In this paper, we present a facial recognition system based on deep learning model. The proposed work is implemented on the embedded system named Raspberry Pi 4. For this, the “Xception convolutional neural network” model is chosen to achieve our emotion system recognition. The registered facial images are taken as input into the classifiers in the embedded system which classifies them into seven facial expressions. For the conduction of the experiment “Fer 2013” data set is used. The proposed model gives an accuracy of 94 % in Graphics Processing Unit” (GPU). After his implementation on the embedded system, according to his limitation against GPU performances, The accuracy is 89% on Raspberry Pi 4. Comparing to other recent works","PeriodicalId":158545,"journal":{"name":"2020 20th International Conference on Sciences and Techniques of Automatic Control and Computer Engineering (STA)","volume":"33 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-12-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128246738","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 : 2020-12-20DOI: 10.1109/STA50679.2020.9329309
K. Bozed, A. Zerek, Amer M. Daeri, Yousef Jaradat
The non-idealities of full-duplex devices of the transceiver chain is not well known despite of the intensive recent research on wireless single-channel full-duplex communications. However, in spite of the use of efficient analog and digital cancellation and suitable physical antenna isolation they turn out to be among main practical reasons for observing residual self-interference. In this paper the implementation and simulation of RF transceiver Duplex Filter and noise isolation improvement using Matlab/SIMULINK environment is done, as well as quantifying the dynamic range of required signal and the reduction of these IF-interference due to the analog-to-digital interface. The Transfer Function calculation results are provided by the help of a White Noise Source and Simulation results of frequency response in the Tx and Rx channels. The simulation and measurement results comparison for new duplexer and the marks and values of frequency response at the critical frequency points have improved isolation due to the effect of optimized external inductor. It has been observed in a full-duplex transceiver that the transmitter power amplifier (PA) produces a nonlinear distortion that is considered to be a significant issue.
{"title":"Modeling and Performance Analysis of the Transceiver Duplex Filter using SIMULINK","authors":"K. Bozed, A. Zerek, Amer M. Daeri, Yousef Jaradat","doi":"10.1109/STA50679.2020.9329309","DOIUrl":"https://doi.org/10.1109/STA50679.2020.9329309","url":null,"abstract":"The non-idealities of full-duplex devices of the transceiver chain is not well known despite of the intensive recent research on wireless single-channel full-duplex communications. However, in spite of the use of efficient analog and digital cancellation and suitable physical antenna isolation they turn out to be among main practical reasons for observing residual self-interference. In this paper the implementation and simulation of RF transceiver Duplex Filter and noise isolation improvement using Matlab/SIMULINK environment is done, as well as quantifying the dynamic range of required signal and the reduction of these IF-interference due to the analog-to-digital interface. The Transfer Function calculation results are provided by the help of a White Noise Source and Simulation results of frequency response in the Tx and Rx channels. The simulation and measurement results comparison for new duplexer and the marks and values of frequency response at the critical frequency points have improved isolation due to the effect of optimized external inductor. It has been observed in a full-duplex transceiver that the transmitter power amplifier (PA) produces a nonlinear distortion that is considered to be a significant issue.","PeriodicalId":158545,"journal":{"name":"2020 20th International Conference on Sciences and Techniques of Automatic Control and Computer Engineering (STA)","volume":"83 ","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-12-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"113988432","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 : 2020-12-20DOI: 10.1109/STA50679.2020.9329292
Chandoul Marwa, S. Othman, H. Sakli
It is estimated that the world's population will be about 9.1 billion by 2050. The UN FAO has reported that food production would need to be increased by approximately 70 percent to feed this increased population. Therefore, to ensure high yields and farm profitability, it is very important to improve agricultural productivity. In this sense, the technology of the Internet of Things (IoT) has become the key road towards novel practice in agriculture. In the agriculture sector, climate change is also a major concern. A solution to completely satisfy the requirements of automated and real-time monitoring of environmental parameters such as humidity, temperature and rain is proposed in this paper. The proposed platform, which collects environmental data (temperature, humidity and rain) over a period of one year was tested on a real farm in Tunisia. The results show that the proposed solution can be used as a reference model to meet the requirements for large-scale agricultural farm calculation, transmission and storage.
{"title":"IoT Based Low-cost Weather Station and Monitoring System for Smart Agriculture","authors":"Chandoul Marwa, S. Othman, H. Sakli","doi":"10.1109/STA50679.2020.9329292","DOIUrl":"https://doi.org/10.1109/STA50679.2020.9329292","url":null,"abstract":"It is estimated that the world's population will be about 9.1 billion by 2050. The UN FAO has reported that food production would need to be increased by approximately 70 percent to feed this increased population. Therefore, to ensure high yields and farm profitability, it is very important to improve agricultural productivity. In this sense, the technology of the Internet of Things (IoT) has become the key road towards novel practice in agriculture. In the agriculture sector, climate change is also a major concern. A solution to completely satisfy the requirements of automated and real-time monitoring of environmental parameters such as humidity, temperature and rain is proposed in this paper. The proposed platform, which collects environmental data (temperature, humidity and rain) over a period of one year was tested on a real farm in Tunisia. The results show that the proposed solution can be used as a reference model to meet the requirements for large-scale agricultural farm calculation, transmission and storage.","PeriodicalId":158545,"journal":{"name":"2020 20th International Conference on Sciences and Techniques of Automatic Control and Computer Engineering (STA)","volume":"21 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-12-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128187061","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 : 2020-12-20DOI: 10.1109/STA50679.2020.9329324
Zeineb Ben Safia, M. Kharrat, M. Allouche, M. Chaabane
In this paper, we deal with a Fault Tolerant Control (FTC) design for an Induction Motor (IM). In spite of the sensor fault and rotor speed variation, this control strategy is able to compensate the fault effect while ensuring the tracking of the desired trajectory, delivered by a Takagi Sugeno (TS) fuzzy reference model. The physical model of the IM is expressed first by a TS fuzzy model and then a TS descriptor observer is developed so that the estimation of the system states and the sensor fault is reached simultaneously. The developed control strategy depends on the Lyapunov theory and H∞ approach, to minimize the disturbances effect. Based on Linear Matrix Inequality (LMI), the controller and the descriptor observer gains are calculated in one phase. At last, simulations have been carried out to show the tracking performance of the designed control strategy.
{"title":"Sensor Fault Tolerant Control For Induction Motor using descriptor approach","authors":"Zeineb Ben Safia, M. Kharrat, M. Allouche, M. Chaabane","doi":"10.1109/STA50679.2020.9329324","DOIUrl":"https://doi.org/10.1109/STA50679.2020.9329324","url":null,"abstract":"In this paper, we deal with a Fault Tolerant Control (FTC) design for an Induction Motor (IM). In spite of the sensor fault and rotor speed variation, this control strategy is able to compensate the fault effect while ensuring the tracking of the desired trajectory, delivered by a Takagi Sugeno (TS) fuzzy reference model. The physical model of the IM is expressed first by a TS fuzzy model and then a TS descriptor observer is developed so that the estimation of the system states and the sensor fault is reached simultaneously. The developed control strategy depends on the Lyapunov theory and H∞ approach, to minimize the disturbances effect. Based on Linear Matrix Inequality (LMI), the controller and the descriptor observer gains are calculated in one phase. At last, simulations have been carried out to show the tracking performance of the designed control strategy.","PeriodicalId":158545,"journal":{"name":"2020 20th International Conference on Sciences and Techniques of Automatic Control and Computer Engineering (STA)","volume":"8 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-12-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133979127","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 : 2020-12-20DOI: 10.1109/STA50679.2020.9329331
M. A. Zeddini, Mourad Turki, Mohamed Faouzi Mimoun
This paper proposes a novel MPPT algorithm using a reinforcement learning (RL) to track the Global Maximum Power Point (GMPP) for photovoltaic (PV) applications. The RL MPPT algorithm was validated by simulation studies under Matlab-simulink for a 2.5 kW PV conversion system based on 5*4 PV modules, a DC/DC converter and a resistive Load. In order to enhance the searching ability of proposed MPPT algorithm, a load and irradiation variations are introduced on simulations tests. In particular, a changing of partial shading condition (PSC) is undertaken to change the position and the value of the GMPP a lot of time for improving the efficiency of the algorithm.
{"title":"Optimization of PV Energy Conversion System Using Reinforcement Learning Algorithm","authors":"M. A. Zeddini, Mourad Turki, Mohamed Faouzi Mimoun","doi":"10.1109/STA50679.2020.9329331","DOIUrl":"https://doi.org/10.1109/STA50679.2020.9329331","url":null,"abstract":"This paper proposes a novel MPPT algorithm using a reinforcement learning (RL) to track the Global Maximum Power Point (GMPP) for photovoltaic (PV) applications. The RL MPPT algorithm was validated by simulation studies under Matlab-simulink for a 2.5 kW PV conversion system based on 5*4 PV modules, a DC/DC converter and a resistive Load. In order to enhance the searching ability of proposed MPPT algorithm, a load and irradiation variations are introduced on simulations tests. In particular, a changing of partial shading condition (PSC) is undertaken to change the position and the value of the GMPP a lot of time for improving the efficiency of the algorithm.","PeriodicalId":158545,"journal":{"name":"2020 20th International Conference on Sciences and Techniques of Automatic Control and Computer Engineering (STA)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-12-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115359118","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 : 2020-12-20DOI: 10.1109/STA50679.2020.9329356
Asma Ounissi, Neila Mezghani Ben Romdhane
The work presented in this paper focuses on the finite-time control of the master-slave manipulators with time-varying delay. First, a nonsingular fast terminal sliding mode control (NFTSMC) is used for this system based on the knowledge of the upper bound of the uncertainties and disturbances. Despite of the presence of the uncertainties, disturbances, load variation and time-varying delay, the controller is robust and present a finite-time convergence. Second, an adaptive nonsingular fast terminal sliding mode control (ANFTSMC) is applied to master-slave manipulators to avoid the knowledge of the upper bound of the uncertainties and disturbances. This controller presents good performance compared to the first one. The two controllers are evaluated in simulation.
{"title":"Adaptive finite-time control of master-slave manipulators with time-varying delay","authors":"Asma Ounissi, Neila Mezghani Ben Romdhane","doi":"10.1109/STA50679.2020.9329356","DOIUrl":"https://doi.org/10.1109/STA50679.2020.9329356","url":null,"abstract":"The work presented in this paper focuses on the finite-time control of the master-slave manipulators with time-varying delay. First, a nonsingular fast terminal sliding mode control (NFTSMC) is used for this system based on the knowledge of the upper bound of the uncertainties and disturbances. Despite of the presence of the uncertainties, disturbances, load variation and time-varying delay, the controller is robust and present a finite-time convergence. Second, an adaptive nonsingular fast terminal sliding mode control (ANFTSMC) is applied to master-slave manipulators to avoid the knowledge of the upper bound of the uncertainties and disturbances. This controller presents good performance compared to the first one. The two controllers are evaluated in simulation.","PeriodicalId":158545,"journal":{"name":"2020 20th International Conference on Sciences and Techniques of Automatic Control and Computer Engineering (STA)","volume":"45 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-12-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124819357","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 : 2020-12-20DOI: 10.1109/STA50679.2020.9329310
E. Leila, S. Othman, H. Sakli
The end of 2019 marked the emergence of a severe acute respiratory syndrome called COVID-19, which created an international public emergency and caused an epidemic. By the moment this paper was written, the number of diagnosed COVID-19 cases around the world reaches more than 30 million. As observed, doctors are at high risk of being infected when they are around patients. To date, physical distancing is the only measure to control the spread of the coronavirus. A statistic made by the World Health Organization (WHO) in September 2020 shows that around 14% of COVID-19 cases reported to WHO are among health workers, illustrating the challenges front-line medical staff face. Robots and drones may be prominent and effective solutions. Besides, robot minimizes person-to-person contact, provide cleaning, education, raise awareness, and support in hospitals and quarantine homes. In this paper, we propose an Internet of Robotic Things System for combating the coronavirus disease pandemic. It collects values of measurable medicals parameters anywhere and anytime and transmits them to the Cloud to be saved and interpreted by a doctor since the healthcare service in the hospital. In hospitals, the robot acting as the intermediary between the physical elements present in the hospital and an information cloud, which saves the values relating to each patient and provides them on-demand via the Internet connection. Keywords: COVID-19, Internet of Things, Sensors, Robot.
{"title":"An Internet of Robotic Things System for combating coronavirus disease pandemic(COVID-19)","authors":"E. Leila, S. Othman, H. Sakli","doi":"10.1109/STA50679.2020.9329310","DOIUrl":"https://doi.org/10.1109/STA50679.2020.9329310","url":null,"abstract":"The end of 2019 marked the emergence of a severe acute respiratory syndrome called COVID-19, which created an international public emergency and caused an epidemic. By the moment this paper was written, the number of diagnosed COVID-19 cases around the world reaches more than 30 million. As observed, doctors are at high risk of being infected when they are around patients. To date, physical distancing is the only measure to control the spread of the coronavirus. A statistic made by the World Health Organization (WHO) in September 2020 shows that around 14% of COVID-19 cases reported to WHO are among health workers, illustrating the challenges front-line medical staff face. Robots and drones may be prominent and effective solutions. Besides, robot minimizes person-to-person contact, provide cleaning, education, raise awareness, and support in hospitals and quarantine homes. In this paper, we propose an Internet of Robotic Things System for combating the coronavirus disease pandemic. It collects values of measurable medicals parameters anywhere and anytime and transmits them to the Cloud to be saved and interpreted by a doctor since the healthcare service in the hospital. In hospitals, the robot acting as the intermediary between the physical elements present in the hospital and an information cloud, which saves the values relating to each patient and provides them on-demand via the Internet connection. Keywords: COVID-19, Internet of Things, Sensors, Robot.","PeriodicalId":158545,"journal":{"name":"2020 20th International Conference on Sciences and Techniques of Automatic Control and Computer Engineering (STA)","volume":"27 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-12-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128393334","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 : 2020-12-20DOI: 10.1109/STA50679.2020.9329330
Ghaith Bouallegue, R. Djemal
Electroencephalogram (EEG) presents a challenge during the classification task using machine learning and deep learning techniques due to the lack or to the low size of available datasets for each specific neurological disorder. Therefore, the use of data augmentation which consists of adding batches of data with patterns quite similar to the training data can offer an interesting solution. Inspired by the successes of the generative adversarial network (GAN) and specifically the Wasserstein GAN (WGAN) version, we propose a deep learning WGAN to generate artificial EEG with features related to each addressed pathogen to approximate the original training dataset. The experimental results demonstrate that using the artificial EEG data generated by our Wasserstein GAN significantly improves the accuracies of the classification models. The implementation was performed using a real dataset dealing with the Autism pathology which is provided by the King Abdulaziz University. Thus, we achieved great results using the presented data augmentation technique applied to the above-mentioned dataset.
脑电图(EEG)在使用机器学习和深度学习技术的分类任务中提出了一个挑战，因为每种特定神经系统疾病的可用数据集缺乏或规模小。因此，使用数据增强(包括添加具有与训练数据非常相似的模式的数据批次)可以提供一个有趣的解决方案。受生成对抗网络(GAN)的成功，特别是Wasserstein GAN (WGAN)版本的启发，我们提出了一种深度学习的WGAN来生成与每个定位病原体相关的特征的人工脑电图，以近似原始训练数据集。实验结果表明，使用我们的Wasserstein GAN生成的人工脑电信号数据显著提高了分类模型的准确性。该实现是使用由阿卜杜勒阿齐兹国王大学提供的处理自闭症病理的真实数据集进行的。因此，我们将所提出的数据增强技术应用于上述数据集，取得了很好的效果。
{"title":"EEG data augmentation using Wasserstein GAN","authors":"Ghaith Bouallegue, R. Djemal","doi":"10.1109/STA50679.2020.9329330","DOIUrl":"https://doi.org/10.1109/STA50679.2020.9329330","url":null,"abstract":"Electroencephalogram (EEG) presents a challenge during the classification task using machine learning and deep learning techniques due to the lack or to the low size of available datasets for each specific neurological disorder. Therefore, the use of data augmentation which consists of adding batches of data with patterns quite similar to the training data can offer an interesting solution. Inspired by the successes of the generative adversarial network (GAN) and specifically the Wasserstein GAN (WGAN) version, we propose a deep learning WGAN to generate artificial EEG with features related to each addressed pathogen to approximate the original training dataset. The experimental results demonstrate that using the artificial EEG data generated by our Wasserstein GAN significantly improves the accuracies of the classification models. The implementation was performed using a real dataset dealing with the Autism pathology which is provided by the King Abdulaziz University. Thus, we achieved great results using the presented data augmentation technique applied to the above-mentioned dataset.","PeriodicalId":158545,"journal":{"name":"2020 20th International Conference on Sciences and Techniques of Automatic Control and Computer Engineering (STA)","volume":"519 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-12-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123124133","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：