Pub Date : 2022-12-22DOI: 10.1109/ISMODE56940.2022.10180423
Y. Jusman, Wikan Tyassari, Wignyo Nindita, Alif Jamil Hussein Harahap, Akbar Maulana Ismail
Identification of Covid-19 use X-ray images to diagnose the level of the covid-19 diseases. The patients can be misdiagnosed due to the similarity between the radiographic images of Covid-19 and pneumonia. Therefore, this research aims to develop automatic screening systems to classify the xray images effectively. Developed Histogram of Oriented Gradients (HOG) algorithm is proposed to be used for features extraction step. The algorithm is developed by enlarging the matrix of extracted features as input to the classification step. The classification step employed three classification algorithms, namely Support Vector Machine (SVM), K-Nearest Neighbor (KNN), and Decision Tree (DT) to classify the image based on the proposed features. The study revealed that the developed HOG algorithm as features extraction method and Medium Gaussian SVM yielded the maximum performance values of 98.28% for accuracy, 97.56% for precision, 97.56% for recall, 98.67% for specificity, and 97.56% for F-score.
{"title":"Developed Histogram of Oriented Gradients-based Feature Extraction for Covid-19 X-Ray Image Classification","authors":"Y. Jusman, Wikan Tyassari, Wignyo Nindita, Alif Jamil Hussein Harahap, Akbar Maulana Ismail","doi":"10.1109/ISMODE56940.2022.10180423","DOIUrl":"https://doi.org/10.1109/ISMODE56940.2022.10180423","url":null,"abstract":"Identification of Covid-19 use X-ray images to diagnose the level of the covid-19 diseases. The patients can be misdiagnosed due to the similarity between the radiographic images of Covid-19 and pneumonia. Therefore, this research aims to develop automatic screening systems to classify the xray images effectively. Developed Histogram of Oriented Gradients (HOG) algorithm is proposed to be used for features extraction step. The algorithm is developed by enlarging the matrix of extracted features as input to the classification step. The classification step employed three classification algorithms, namely Support Vector Machine (SVM), K-Nearest Neighbor (KNN), and Decision Tree (DT) to classify the image based on the proposed features. The study revealed that the developed HOG algorithm as features extraction method and Medium Gaussian SVM yielded the maximum performance values of 98.28% for accuracy, 97.56% for precision, 97.56% for recall, 98.67% for specificity, and 97.56% for F-score.","PeriodicalId":335247,"journal":{"name":"2022 2nd International Seminar on Machine Learning, Optimization, and Data Science (ISMODE)","volume":"161 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123533021","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 : 2022-12-22DOI: 10.1109/ISMODE56940.2022.10180990
Ali Alarjani, Teg Alam, A. Kineber
Forecasting models are essential for economic development and making appropriate policy decisions. The purpose of this study is to forecast crude oil production in Saudi Arabia for the following year. Our study is aimed at predicting Saudi Arabia’s crude oil production using Artificial Neural Networks (ANN), Holt-Winters Exponential Smoothing (HW), and Autoregressive Integrated Moving Averages (ARIMA). Based on 1993-2022 crude oil production (million barrels per day) data, this study applies statistical analysis to forecast time series data based on said models over a period. The study also analyzes the forecast model’s accuracy using a variety of measures. As a result of the analysis, this study found that ANNs are the most effective at predicting crude oil production. Thus, among other models analyzed in this study, the ANN model can accurately predict Saudi Arabia’s crude oil production in the future. In addition, the study aims to clarify the current situation of crude oil production in the kingdom. Researchers will be able to better understand crude oil production forecasts as a result of this study. This study can also provide guidance for developing a strategic plan for government entities.
{"title":"Modeling and Predicting Saudi Crude Oil Production Using Artificial Neural Networks (ANN) and Some Others Predictive Techniques","authors":"Ali Alarjani, Teg Alam, A. Kineber","doi":"10.1109/ISMODE56940.2022.10180990","DOIUrl":"https://doi.org/10.1109/ISMODE56940.2022.10180990","url":null,"abstract":"Forecasting models are essential for economic development and making appropriate policy decisions. The purpose of this study is to forecast crude oil production in Saudi Arabia for the following year. Our study is aimed at predicting Saudi Arabia’s crude oil production using Artificial Neural Networks (ANN), Holt-Winters Exponential Smoothing (HW), and Autoregressive Integrated Moving Averages (ARIMA). Based on 1993-2022 crude oil production (million barrels per day) data, this study applies statistical analysis to forecast time series data based on said models over a period. The study also analyzes the forecast model’s accuracy using a variety of measures. As a result of the analysis, this study found that ANNs are the most effective at predicting crude oil production. Thus, among other models analyzed in this study, the ANN model can accurately predict Saudi Arabia’s crude oil production in the future. In addition, the study aims to clarify the current situation of crude oil production in the kingdom. Researchers will be able to better understand crude oil production forecasts as a result of this study. This study can also provide guidance for developing a strategic plan for government entities.","PeriodicalId":335247,"journal":{"name":"2022 2nd International Seminar on Machine Learning, Optimization, and Data Science (ISMODE)","volume":"12 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116330732","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 : 2022-12-22DOI: 10.1109/ISMODE56940.2022.10180915
Daryl B. Valdez, Rey Anthony G. Godmalin, Allan Josephus M. Bunga
Natural disasters are destructive forces that greatly affect all people around the world. In the Philippines, earthquakes, tropical cyclones, and floods are some of the most frequent disasters that struck the country in recent years. Several studies utilizing technology have been conducted with varying degrees of success to reduce the impact of these uncontrollable events. Different from others, this paper investigates the use of a Deep Learning model deployed in a Raspberry Pi 3b for on-the-ground, real-time, automated disaster recognition and monitoring. It aims to empower emergency responders and people in the community to easily detect disasters as they happen in real-time, reducing the loss of life and damage to property. To this end, a novel low-cost monitoring system is proposed. Experiments and a survey made to emergency responders were conducted to validate the system’s feasibility and acceptability. Results revealed that the proposed system detects disasters with a high degree of performance. Also, it utilizes a low CPU and memory footprint while achieving seven frames per second processing rate during disaster recognition. In addition, the respondents find the system clear, helpful, innovative, and easy to use. Hence, the system is capable of recognizing disasters in real-time, proving acceptable and beneficial to people in the community.
{"title":"Vision-based Real-Time Disaster Recognition Monitoring System using Raspberry Pi and Deep Learning Model","authors":"Daryl B. Valdez, Rey Anthony G. Godmalin, Allan Josephus M. Bunga","doi":"10.1109/ISMODE56940.2022.10180915","DOIUrl":"https://doi.org/10.1109/ISMODE56940.2022.10180915","url":null,"abstract":"Natural disasters are destructive forces that greatly affect all people around the world. In the Philippines, earthquakes, tropical cyclones, and floods are some of the most frequent disasters that struck the country in recent years. Several studies utilizing technology have been conducted with varying degrees of success to reduce the impact of these uncontrollable events. Different from others, this paper investigates the use of a Deep Learning model deployed in a Raspberry Pi 3b for on-the-ground, real-time, automated disaster recognition and monitoring. It aims to empower emergency responders and people in the community to easily detect disasters as they happen in real-time, reducing the loss of life and damage to property. To this end, a novel low-cost monitoring system is proposed. Experiments and a survey made to emergency responders were conducted to validate the system’s feasibility and acceptability. Results revealed that the proposed system detects disasters with a high degree of performance. Also, it utilizes a low CPU and memory footprint while achieving seven frames per second processing rate during disaster recognition. In addition, the respondents find the system clear, helpful, innovative, and easy to use. Hence, the system is capable of recognizing disasters in real-time, proving acceptable and beneficial to people in the community.","PeriodicalId":335247,"journal":{"name":"2022 2nd International Seminar on Machine Learning, Optimization, and Data Science (ISMODE)","volume":"39 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124822343","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 : 2022-12-22DOI: 10.1109/ISMODE56940.2022.10181007
A. N. Handayani, Ferina Ayu Pusparani, D. Lestari, I. M. Wirawan, Osamu Fukuda, Aqil Aqthobirrobbany
Enhancing maritime patrol to strengthen state border security has shown numerous interests over the year. Marine patrol is crucial in raising maritime awareness and surveying what is happening in the vast Indonesian sea area. Ship detection and identification are essential for marine patrol dealing with maritime traffic, sea border activity, and illegal fishery. Because of that, object detection integrated with the autonomous surface vehicle, like robot boats, is an advantageous method used in marine patrol. The robot boat used an object detection algorithm processed by Jetson Nano to determine its navigation. Preliminary experiments are conducted to verify if the proposed method can recognize an object and patrol the surrounding area in real-time using the integrated surface robot.
{"title":"Robot Boat Prototype System Based on Image Processing for Maritime Patrol Area","authors":"A. N. Handayani, Ferina Ayu Pusparani, D. Lestari, I. M. Wirawan, Osamu Fukuda, Aqil Aqthobirrobbany","doi":"10.1109/ISMODE56940.2022.10181007","DOIUrl":"https://doi.org/10.1109/ISMODE56940.2022.10181007","url":null,"abstract":"Enhancing maritime patrol to strengthen state border security has shown numerous interests over the year. Marine patrol is crucial in raising maritime awareness and surveying what is happening in the vast Indonesian sea area. Ship detection and identification are essential for marine patrol dealing with maritime traffic, sea border activity, and illegal fishery. Because of that, object detection integrated with the autonomous surface vehicle, like robot boats, is an advantageous method used in marine patrol. The robot boat used an object detection algorithm processed by Jetson Nano to determine its navigation. Preliminary experiments are conducted to verify if the proposed method can recognize an object and patrol the surrounding area in real-time using the integrated surface robot.","PeriodicalId":335247,"journal":{"name":"2022 2nd International Seminar on Machine Learning, Optimization, and Data Science (ISMODE)","volume":"32 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116687086","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 : 2022-12-22DOI: 10.1109/ISMODE56940.2022.10180931
Machrus Ali, M. Djalal, Hidayatul Nurohmah, Rukslin
A Proportional Integral Derivative (PID) controller in a synchronous motor is widely used because of its simple structure, robustness, strength and ease of use. The use of a PID controller requires proper parameter settings for optimal performance on the motor. The solution often used is the trial-error method to determine the correct parameters for the PID, but the results obtained do not make the PID controller optimal. Recently there have been many studies to optimize PID controllers wrong with intelligent methods. For this reason, this research will use the Craziness Particle Swarm Optimization (CRPSO) optimization method to optimize and determine the proper parameters of the PID. The CRPSO method is a method that provides an innovation to the velocity function of the particles distributed in the PSO method. From the simulation results, CRPSO performance is more optimal than PSO. From the correct PID parameter tuning results, a minimum overshoot response is obtained with several speed variations. In addition, an increase was also obtained in PMSM starting torque using CRPSO.
{"title":"Intelligent Optimization Using Craziness Particle Swarm on Permanent Magnet Synchronous Motor","authors":"Machrus Ali, M. Djalal, Hidayatul Nurohmah, Rukslin","doi":"10.1109/ISMODE56940.2022.10180931","DOIUrl":"https://doi.org/10.1109/ISMODE56940.2022.10180931","url":null,"abstract":"A Proportional Integral Derivative (PID) controller in a synchronous motor is widely used because of its simple structure, robustness, strength and ease of use. The use of a PID controller requires proper parameter settings for optimal performance on the motor. The solution often used is the trial-error method to determine the correct parameters for the PID, but the results obtained do not make the PID controller optimal. Recently there have been many studies to optimize PID controllers wrong with intelligent methods. For this reason, this research will use the Craziness Particle Swarm Optimization (CRPSO) optimization method to optimize and determine the proper parameters of the PID. The CRPSO method is a method that provides an innovation to the velocity function of the particles distributed in the PSO method. From the simulation results, CRPSO performance is more optimal than PSO. From the correct PID parameter tuning results, a minimum overshoot response is obtained with several speed variations. In addition, an increase was also obtained in PMSM starting torque using CRPSO.","PeriodicalId":335247,"journal":{"name":"2022 2nd International Seminar on Machine Learning, Optimization, and Data Science (ISMODE)","volume":"160 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121189293","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 : 2022-12-22DOI: 10.1109/ISMODE56940.2022.10181009
Fajar Dwi Cahyoko, Sutikno, Purhadi
Numerical Weather Prediction is a weather forecasting method that is translated into a system of mathematical equations that are solved by numerical methods. The transformation of the basic theory of NWP into computer code still produces errors. To reduce errors and increase the accuracy of the prediction results of the NWP model, statistical postprocessing can be performed using the Model Output Statistics (MOS) method. The use of model output statistics for weather prediction still has a deficiency, namely, it still produces high bias. To increase the accuracy of the prediction model, it can use the ensemble model output statistics (EMOS). This approach is set out from the ensemble prediction system (EPS) which has an understanding as a model consisting of a combination of two or more single prediction models that are verified at the same time. This technique yields probabilistic forecasts that take the form of Gaussian predictive probability density functions (PDFs) for continuous weather variables. The ensemble members in this study consist of prediction from PLS, PCR, and Ridge Regression. In these performances, EMOS offers predictive PDF and CDF from an ensemble forecast of a continuous weather variable, but it is not considered spatial correlation. For the training period over 20,30 and 40 days, EMOS temperature forecast at 3 sites into good and fair ones. Based on weather prediction assessment indicators like RMSE and CRPS, EMOS is better than raw ensemble in terms of accuracy and precision.
{"title":"The Usage of Ensemble Model Output Statistics for Calibration and Short-term Weather Forecast","authors":"Fajar Dwi Cahyoko, Sutikno, Purhadi","doi":"10.1109/ISMODE56940.2022.10181009","DOIUrl":"https://doi.org/10.1109/ISMODE56940.2022.10181009","url":null,"abstract":"Numerical Weather Prediction is a weather forecasting method that is translated into a system of mathematical equations that are solved by numerical methods. The transformation of the basic theory of NWP into computer code still produces errors. To reduce errors and increase the accuracy of the prediction results of the NWP model, statistical postprocessing can be performed using the Model Output Statistics (MOS) method. The use of model output statistics for weather prediction still has a deficiency, namely, it still produces high bias. To increase the accuracy of the prediction model, it can use the ensemble model output statistics (EMOS). This approach is set out from the ensemble prediction system (EPS) which has an understanding as a model consisting of a combination of two or more single prediction models that are verified at the same time. This technique yields probabilistic forecasts that take the form of Gaussian predictive probability density functions (PDFs) for continuous weather variables. The ensemble members in this study consist of prediction from PLS, PCR, and Ridge Regression. In these performances, EMOS offers predictive PDF and CDF from an ensemble forecast of a continuous weather variable, but it is not considered spatial correlation. For the training period over 20,30 and 40 days, EMOS temperature forecast at 3 sites into good and fair ones. Based on weather prediction assessment indicators like RMSE and CRPS, EMOS is better than raw ensemble in terms of accuracy and precision.","PeriodicalId":335247,"journal":{"name":"2022 2nd International Seminar on Machine Learning, Optimization, and Data Science (ISMODE)","volume":"17 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126737564","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 : 2022-12-22DOI: 10.1109/ISMODE56940.2022.10180926
P. Agbedanu, R. Musabe, James Rwigema, Ignace Gatare, Yanis Pavlidis
Intrusion Detection Systems (IDSs) in traditional computing systems have played a significant role in detecting and preventing cyber-attacks. Unsurprisingly, the same technology is used to detect and prevent cyber attacks in Internet of Things (IoT) environments. However, due to the computational constraints of IoT devices, traditional computing-based IDS is challenging to deploy on IoT devices. Moreover, IDS for IoT environments should have high classification performance, low complexity models, and small model sizes. Despite numerous advances in IoT-based intrusion detection, developing models that achieve high classification performance while being less complex and smaller in size remains difficult. This study proposes a novel IDS for resource-constrained devices like IoT systems by using a blend of incremental principal component analysis (IPCA) and Self Adjusting Memory KNN (SAM-KNN) to develop a lightweight machine learning model to detect intrusions in IoT systems. The proposed system was deployed on a Raspberry Pi Model B, representing a resource-constrained device, and evaluated using the UNSW-NB15 dataset. The experimental results show a superior accuracy of 98.91%, a memory overhead of 1.4%, 1.6% and 2% overhead for CPU and energy, respectively.
{"title":"IPCA-SAMKNN: A Novel Network IDS for Resource Constrained Devices","authors":"P. Agbedanu, R. Musabe, James Rwigema, Ignace Gatare, Yanis Pavlidis","doi":"10.1109/ISMODE56940.2022.10180926","DOIUrl":"https://doi.org/10.1109/ISMODE56940.2022.10180926","url":null,"abstract":"Intrusion Detection Systems (IDSs) in traditional computing systems have played a significant role in detecting and preventing cyber-attacks. Unsurprisingly, the same technology is used to detect and prevent cyber attacks in Internet of Things (IoT) environments. However, due to the computational constraints of IoT devices, traditional computing-based IDS is challenging to deploy on IoT devices. Moreover, IDS for IoT environments should have high classification performance, low complexity models, and small model sizes. Despite numerous advances in IoT-based intrusion detection, developing models that achieve high classification performance while being less complex and smaller in size remains difficult. This study proposes a novel IDS for resource-constrained devices like IoT systems by using a blend of incremental principal component analysis (IPCA) and Self Adjusting Memory KNN (SAM-KNN) to develop a lightweight machine learning model to detect intrusions in IoT systems. The proposed system was deployed on a Raspberry Pi Model B, representing a resource-constrained device, and evaluated using the UNSW-NB15 dataset. The experimental results show a superior accuracy of 98.91%, a memory overhead of 1.4%, 1.6% and 2% overhead for CPU and energy, respectively.","PeriodicalId":335247,"journal":{"name":"2022 2nd International Seminar on Machine Learning, Optimization, and Data Science (ISMODE)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130733174","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 : 2022-12-22DOI: 10.1109/ISMODE56940.2022.10181008
Moh. Afandy, Muh. Alif Nur, Abdul Haris Mubarak
This study presents the results of the IOT-based 3-phase electric motor monitoring system design in the Nickel smelting industry. Conventional measurement methods have a high level of risk and are very dangerous for field workers. The use of automatic measurement methods is considered better because the accuracy of measurement results can reach 98% with measurement data that can be well documented in the storage system. The use of voltage and current sensors in this design uses Current Transformer (CT) and Potential Transformer (PT) which are industry standard devices. The measurement signal obtained from the CT and PT components is processed using a signal conditioning circuit. Setting the AC to DC voltage in the signal conditioning circuit is the initial stage in signal management, then the results are amplified using signal amplification, which is needed to increase the accuracy of readings in the control circuit using OP-AMP amplification. The results of reading the current, voltage, and power measurement data from a 3-phase electric motor will be displayed on the HMI using the website. The measurement results are in the form of data stored in database storage to facilitate the evaluation process of daily, weekly, and even annual power usage. From these results, the power of using a 3-phase electric motor with an additional generator load of 3212, 563W is obtained. In processing the value of the current measurement results in each phase, the difference in the value of the phase current is obtained, namely 5.11A, 4.36A, 4.06A. Voltage measurements in each phase are also obtained through data processing voltages of 232.13 V, 242.03 V, 238.16 V for the R, S, and T phases. Finally, the monitoring system design this time can be implemented.
{"title":"Power Monitoring System for 3-Phase Electric Motors Using IoT-Based Current Transformers and Potential Transformers","authors":"Moh. Afandy, Muh. Alif Nur, Abdul Haris Mubarak","doi":"10.1109/ISMODE56940.2022.10181008","DOIUrl":"https://doi.org/10.1109/ISMODE56940.2022.10181008","url":null,"abstract":"This study presents the results of the IOT-based 3-phase electric motor monitoring system design in the Nickel smelting industry. Conventional measurement methods have a high level of risk and are very dangerous for field workers. The use of automatic measurement methods is considered better because the accuracy of measurement results can reach 98% with measurement data that can be well documented in the storage system. The use of voltage and current sensors in this design uses Current Transformer (CT) and Potential Transformer (PT) which are industry standard devices. The measurement signal obtained from the CT and PT components is processed using a signal conditioning circuit. Setting the AC to DC voltage in the signal conditioning circuit is the initial stage in signal management, then the results are amplified using signal amplification, which is needed to increase the accuracy of readings in the control circuit using OP-AMP amplification. The results of reading the current, voltage, and power measurement data from a 3-phase electric motor will be displayed on the HMI using the website. The measurement results are in the form of data stored in database storage to facilitate the evaluation process of daily, weekly, and even annual power usage. From these results, the power of using a 3-phase electric motor with an additional generator load of 3212, 563W is obtained. In processing the value of the current measurement results in each phase, the difference in the value of the phase current is obtained, namely 5.11A, 4.36A, 4.06A. Voltage measurements in each phase are also obtained through data processing voltages of 232.13 V, 242.03 V, 238.16 V for the R, S, and T phases. Finally, the monitoring system design this time can be implemented.","PeriodicalId":335247,"journal":{"name":"2022 2nd International Seminar on Machine Learning, Optimization, and Data Science (ISMODE)","volume":"2007 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131328333","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 : 2022-12-22DOI: 10.1109/ISMODE56940.2022.10180952
J. Gamazo-Real, Raúl Torres Fernández, Adrián Murillo Armas
The vast increase in connected Internet of Things (IoT) devices have revolutionised how data are processed. This fact, coupled with the current trend from cloud to edge computing paradigms, has resulted in the need for efficient and reliable data processing near to data sources using resource-constrained devices. In this article, low-cost edge-IoT architectures are implemented to deploy lightweight Machine Learning (ML) models for air quality estimation, such as Polynomial Regression and Artificial Neural Networks (ANN). ML models are deployed in wireless centralised and distributed parallel architectures with common modules such as sensor fusion for luminosity, temperature, humidity, CO2, and other gases. The centralised architecture uses a Graphic Processing Unit (GPU) and the Message Queuing Telemetry Transport (MQTT) protocol, but low-performance processing devices and the Message Passing Interface (MPI) protocol are used in the distributed one. The training and testing of models are attained with appropriate datasets obtained from multiple peak, step, and transient test cases for each air quality parameter. The results for temperature forecasting, and similar ones for other parameters, supports that the distributed parallel architecture could achieve a slightly better estimation metrics and a better performance in power consumption compared to the centralised architecture despite using low-cost general purpose devices.
{"title":"Estimation of Air Quality Parameters using Lightweight Machine Learning on Low-cost Edge-IoT Architectures","authors":"J. Gamazo-Real, Raúl Torres Fernández, Adrián Murillo Armas","doi":"10.1109/ISMODE56940.2022.10180952","DOIUrl":"https://doi.org/10.1109/ISMODE56940.2022.10180952","url":null,"abstract":"The vast increase in connected Internet of Things (IoT) devices have revolutionised how data are processed. This fact, coupled with the current trend from cloud to edge computing paradigms, has resulted in the need for efficient and reliable data processing near to data sources using resource-constrained devices. In this article, low-cost edge-IoT architectures are implemented to deploy lightweight Machine Learning (ML) models for air quality estimation, such as Polynomial Regression and Artificial Neural Networks (ANN). ML models are deployed in wireless centralised and distributed parallel architectures with common modules such as sensor fusion for luminosity, temperature, humidity, CO2, and other gases. The centralised architecture uses a Graphic Processing Unit (GPU) and the Message Queuing Telemetry Transport (MQTT) protocol, but low-performance processing devices and the Message Passing Interface (MPI) protocol are used in the distributed one. The training and testing of models are attained with appropriate datasets obtained from multiple peak, step, and transient test cases for each air quality parameter. The results for temperature forecasting, and similar ones for other parameters, supports that the distributed parallel architecture could achieve a slightly better estimation metrics and a better performance in power consumption compared to the centralised architecture despite using low-cost general purpose devices.","PeriodicalId":335247,"journal":{"name":"2022 2nd International Seminar on Machine Learning, Optimization, and Data Science (ISMODE)","volume":"23 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130851361","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 : 2022-12-22DOI: 10.1109/ISMODE56940.2022.10180411
Rizky Naufal Perdana, Budhi Irawan, C. Setianingsih, Dian Rezky Wulandari, Ivan Satrio Pamungkas, Fajri Nurfauzan, Adinda Ophelia Putri Sakinah, Muhammad Raihan Ramadhan
In the current era, technology is developing very rapidly, especially in the field of image processing, technological developments can help and facilitate human work. The purpose of image processing is to learn how to process images to detect objects. This project has the aim of implementing real face detection based on image processing in the smart door design. The physiological motion detection system can recognize the difference between real faces and photo imitations based on facial reflexes in the eyes and mouth. The methods used for face detection are Histogram Oriented Gradient and Haar-Cascade, motion detection of facial reflexes using the Support Vector Machine. The result of this project concludes that the smart door system with physiological motion detection that has been designed successfully performs real face detection, the average accuracy rate is 93.5% and photo imitation faces have an average of 90.7% based on eye reflex motion detection. and mouth.
{"title":"Design of Smartdoor for Live Face Detection Based on Image Processing Using Physiological Motion Detection","authors":"Rizky Naufal Perdana, Budhi Irawan, C. Setianingsih, Dian Rezky Wulandari, Ivan Satrio Pamungkas, Fajri Nurfauzan, Adinda Ophelia Putri Sakinah, Muhammad Raihan Ramadhan","doi":"10.1109/ISMODE56940.2022.10180411","DOIUrl":"https://doi.org/10.1109/ISMODE56940.2022.10180411","url":null,"abstract":"In the current era, technology is developing very rapidly, especially in the field of image processing, technological developments can help and facilitate human work. The purpose of image processing is to learn how to process images to detect objects. This project has the aim of implementing real face detection based on image processing in the smart door design. The physiological motion detection system can recognize the difference between real faces and photo imitations based on facial reflexes in the eyes and mouth. The methods used for face detection are Histogram Oriented Gradient and Haar-Cascade, motion detection of facial reflexes using the Support Vector Machine. The result of this project concludes that the smart door system with physiological motion detection that has been designed successfully performs real face detection, the average accuracy rate is 93.5% and photo imitation faces have an average of 90.7% based on eye reflex motion detection. and mouth.","PeriodicalId":335247,"journal":{"name":"2022 2nd International Seminar on Machine Learning, Optimization, and Data Science (ISMODE)","volume":"15 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134629702","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：