Pub Date : 2023-05-09DOI: 10.1109/ECTI-CON58255.2023.10153275
Rom Khanpeth, M. Kongpoon, A. Thanachayanont
This paper describes the design of a low-power temperature sensor in a 0.18-μm CMOS technology. The proposed temperature sensor employs the so-called “dynamic threshold MOS (DTMOS)" diode-connected transistors as the temperature sensing devices. Process spread of the MOSFET threshold voltage is compensated by using the 2-transistor (2T) voltage reference to generate the bias current sources. A charge-balancing delta-sigma $(Delta Sigma)$ analog-to-digital converter (ADC) is used to obtain the digital representation of temperature values. The DTMOS temperature sensor core and the ADC operate with 1 V power supply voltages. The ADC operates with a 64- kHz clock frequency and each temperature conversion time is 32ms. After a single-point temperature trimming and a linear fit, the proposed circuit achieves a maximum inaccuracy of ±0.25°C (3σ) across all process corners and the temperature range of −20°C to 85°C, while consuming 8.1 μW.
{"title":"A DTMOS-based temperature sensor with an inaccuracy of ±0.25°C (3σ) from −20°C to 85°C","authors":"Rom Khanpeth, M. Kongpoon, A. Thanachayanont","doi":"10.1109/ECTI-CON58255.2023.10153275","DOIUrl":"https://doi.org/10.1109/ECTI-CON58255.2023.10153275","url":null,"abstract":"This paper describes the design of a low-power temperature sensor in a 0.18-μm CMOS technology. The proposed temperature sensor employs the so-called “dynamic threshold MOS (DTMOS)\" diode-connected transistors as the temperature sensing devices. Process spread of the MOSFET threshold voltage is compensated by using the 2-transistor (2T) voltage reference to generate the bias current sources. A charge-balancing delta-sigma $(Delta Sigma)$ analog-to-digital converter (ADC) is used to obtain the digital representation of temperature values. The DTMOS temperature sensor core and the ADC operate with 1 V power supply voltages. The ADC operates with a 64- kHz clock frequency and each temperature conversion time is 32ms. After a single-point temperature trimming and a linear fit, the proposed circuit achieves a maximum inaccuracy of ±0.25°C (3σ) across all process corners and the temperature range of −20°C to 85°C, while consuming 8.1 μW.","PeriodicalId":340768,"journal":{"name":"2023 20th International Conference on Electrical Engineering/Electronics, Computer, Telecommunications and Information Technology (ECTI-CON)","volume":"11 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133187440","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}
This research presents construction of noncontact warning gloves for technician who works in electrical system in the building. The insulation gloves were used. The warning circuit by magnetic field detector is installed in the gloves. So the circuit can detect during non-touch. IC 74HC14 is applied in warning circuit. The warning displays by LED and sound. The results show that warning gloves can detect the flow current in conductor at the range of 0.5-5 A. which this range accord to standard safety. The implement are tested with 5 building electrical technicians in two sectors; feature and satisfactions. It is found that the suitable feature is very good level at the average of 4.55. The satisfaction of respondents is very good level also at the average of 4.60. In conclusion, the warning gloves can use effectively.
{"title":"A Construction of Non-touch Warning Gloves for Performance of Building Electrical Technician","authors":"Tharathip Phurahong, Settasuk Kwuanphet, Theerawat Tibeye","doi":"10.1109/ECTI-CON58255.2023.10153295","DOIUrl":"https://doi.org/10.1109/ECTI-CON58255.2023.10153295","url":null,"abstract":"This research presents construction of noncontact warning gloves for technician who works in electrical system in the building. The insulation gloves were used. The warning circuit by magnetic field detector is installed in the gloves. So the circuit can detect during non-touch. IC 74HC14 is applied in warning circuit. The warning displays by LED and sound. The results show that warning gloves can detect the flow current in conductor at the range of 0.5-5 A. which this range accord to standard safety. The implement are tested with 5 building electrical technicians in two sectors; feature and satisfactions. It is found that the suitable feature is very good level at the average of 4.55. The satisfaction of respondents is very good level also at the average of 4.60. In conclusion, the warning gloves can use effectively.","PeriodicalId":340768,"journal":{"name":"2023 20th International Conference on Electrical Engineering/Electronics, Computer, Telecommunications and Information Technology (ECTI-CON)","volume":"85 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115824928","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 : 2023-05-09DOI: 10.1109/ECTI-CON58255.2023.10153146
J. Thanawiparat, T. Kasetkasem, T. Patrapornnant, S. Patarapuwadol
In this paper, we propose a new approach for the detection and classification of rice diseases that combines the use of deep learning and level set methods. Our approach aims to improve the performance of rice disease classification and object detection by using an efficient image processing technique with deep learning. The proposed method will be evaluated based on Mean Average Precision (mAP), Accuracy, Precision, Recall, and F1-score. The results indicate that our proposed method improved the performance for most of the 16 classes of rice diseases and also it was able to improve performance even when using a small amount of labeled data. We also found that our proposed algorithm produced segmentation maps with significantly smaller computational time and the use of bias field estimation helped in the segmentation of complicated foreground objects in images by modifying unnecessary details. Overall, this research has the potential to be a valuable tool in the field of agriculture, where early and accurate detection of rice diseases is crucial for preventing the spread of the disease and maintaining crop yields.
{"title":"Rice Diseases Detection and Localization with Only Image-Level Disease Training Labels","authors":"J. Thanawiparat, T. Kasetkasem, T. Patrapornnant, S. Patarapuwadol","doi":"10.1109/ECTI-CON58255.2023.10153146","DOIUrl":"https://doi.org/10.1109/ECTI-CON58255.2023.10153146","url":null,"abstract":"In this paper, we propose a new approach for the detection and classification of rice diseases that combines the use of deep learning and level set methods. Our approach aims to improve the performance of rice disease classification and object detection by using an efficient image processing technique with deep learning. The proposed method will be evaluated based on Mean Average Precision (mAP), Accuracy, Precision, Recall, and F1-score. The results indicate that our proposed method improved the performance for most of the 16 classes of rice diseases and also it was able to improve performance even when using a small amount of labeled data. We also found that our proposed algorithm produced segmentation maps with significantly smaller computational time and the use of bias field estimation helped in the segmentation of complicated foreground objects in images by modifying unnecessary details. Overall, this research has the potential to be a valuable tool in the field of agriculture, where early and accurate detection of rice diseases is crucial for preventing the spread of the disease and maintaining crop yields.","PeriodicalId":340768,"journal":{"name":"2023 20th International Conference on Electrical Engineering/Electronics, Computer, Telecommunications and Information Technology (ECTI-CON)","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123929201","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 : 2023-05-09DOI: 10.1109/ECTI-CON58255.2023.10153281
Chanjinda Taesuwan, Sakkrapop Khuangsatung, N. Wichaipanich, K. Hozumi
This paper presents the vertical plasma drifts velocity of the F2-layer (Vz-hmF2) compared with the IRI-2016 model. The Vz-hmF2 data was obtained from the equatorial ionosonde station at Chumphon (Lat. 10.72°N, Long. 99.37°E, Dip 3.0°N), Thailand. The Vz-hmF2 during low solar activity in 2010 was analyzed the daily and monthly variations between both the observed and the modelled. The results show the IRI-2016 model overestimates the observed Vz during noontime to midnight while it underestimates from pre-sunrise to noontime hours. The observed Vz changes from -60 to 110 m/s and the model varies between $pm 20m/s$. This paper will be helpful the improvement of the IRI model in the future.
{"title":"Vertical plasma drift velocity of F2 layer inferred from ionosonde compared with IRI-2016 Model Over Chumphon, Thailand","authors":"Chanjinda Taesuwan, Sakkrapop Khuangsatung, N. Wichaipanich, K. Hozumi","doi":"10.1109/ECTI-CON58255.2023.10153281","DOIUrl":"https://doi.org/10.1109/ECTI-CON58255.2023.10153281","url":null,"abstract":"This paper presents the vertical plasma drifts velocity of the F2-layer (Vz-hmF2) compared with the IRI-2016 model. The Vz-hmF2 data was obtained from the equatorial ionosonde station at Chumphon (Lat. 10.72°N, Long. 99.37°E, Dip 3.0°N), Thailand. The Vz-hmF2 during low solar activity in 2010 was analyzed the daily and monthly variations between both the observed and the modelled. The results show the IRI-2016 model overestimates the observed Vz during noontime to midnight while it underestimates from pre-sunrise to noontime hours. The observed Vz changes from -60 to 110 m/s and the model varies between $pm 20m/s$. This paper will be helpful the improvement of the IRI model in the future.","PeriodicalId":340768,"journal":{"name":"2023 20th International Conference on Electrical Engineering/Electronics, Computer, Telecommunications and Information Technology (ECTI-CON)","volume":"20 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126136192","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}
Train position and velocity determination are of great importance for railway systems since uninterrupted data of train position and velocity is required for the signaling system to manage the traffic and maintain the safety of railway network. Hence specific train positioning system using a satellite network is developed. Due to very high investment, the satellite train positioning system is available in only some countries and limited classes of vehicles. Train localization using GPS data is an alternative, and a dead reckoning system is integrated to handle intermittent GPS data. Since the dynamic model of the train is unavailable, the proposed estimator is created by using a PID controller and double integrator. The developed system is installed and tested on a commuter train in Bangkok and the device can provide the train position and velocity in real-time without interruption.
{"title":"Development of train positioning and dead reckoning system using GPS","authors":"Wisute Ongcunaruck, Witthawas Pongyart, Kamphon Rattanakijkamol","doi":"10.1109/ECTI-CON58255.2023.10153136","DOIUrl":"https://doi.org/10.1109/ECTI-CON58255.2023.10153136","url":null,"abstract":"Train position and velocity determination are of great importance for railway systems since uninterrupted data of train position and velocity is required for the signaling system to manage the traffic and maintain the safety of railway network. Hence specific train positioning system using a satellite network is developed. Due to very high investment, the satellite train positioning system is available in only some countries and limited classes of vehicles. Train localization using GPS data is an alternative, and a dead reckoning system is integrated to handle intermittent GPS data. Since the dynamic model of the train is unavailable, the proposed estimator is created by using a PID controller and double integrator. The developed system is installed and tested on a commuter train in Bangkok and the device can provide the train position and velocity in real-time without interruption.","PeriodicalId":340768,"journal":{"name":"2023 20th International Conference on Electrical Engineering/Electronics, Computer, Telecommunications and Information Technology (ECTI-CON)","volume":"2675 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128454046","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 : 2023-05-09DOI: 10.1109/ECTI-CON58255.2023.10153162
Thanasit Rithanasophon, C. Wannaboon
Fingerprint-based indoor positioning systems are simple and widely used to determine the location of a device inside a building or other enclosed area. However, the accuracy and reliability are still a major concern due to the turbulence in the environment and the presence of noise in the data. This paper presents a machine learning integrated with Kalman filter approach for improving the accuracy of fingerprint-based indoor positioning systems. The proposed approach combines the power of machine learning techniques for feature extraction and classification with the noise-filtering capabilities of the Kalman filter. Implementation is achieved by a real-world dataset collected from multiple Bluetooth low energy access points. The experiment results indicate that the proposed approach significantly improves the accuracy of fingerprint-based indoor positioning compared to traditional machine learning approaches. This study also offers a potential of cost-effective and high accuracy algorithm for the indoor positioning applications.
{"title":"Integration of Machine Learning and Kalman Filter Approach for Fingerprint Indoor Positioning","authors":"Thanasit Rithanasophon, C. Wannaboon","doi":"10.1109/ECTI-CON58255.2023.10153162","DOIUrl":"https://doi.org/10.1109/ECTI-CON58255.2023.10153162","url":null,"abstract":"Fingerprint-based indoor positioning systems are simple and widely used to determine the location of a device inside a building or other enclosed area. However, the accuracy and reliability are still a major concern due to the turbulence in the environment and the presence of noise in the data. This paper presents a machine learning integrated with Kalman filter approach for improving the accuracy of fingerprint-based indoor positioning systems. The proposed approach combines the power of machine learning techniques for feature extraction and classification with the noise-filtering capabilities of the Kalman filter. Implementation is achieved by a real-world dataset collected from multiple Bluetooth low energy access points. The experiment results indicate that the proposed approach significantly improves the accuracy of fingerprint-based indoor positioning compared to traditional machine learning approaches. This study also offers a potential of cost-effective and high accuracy algorithm for the indoor positioning applications.","PeriodicalId":340768,"journal":{"name":"2023 20th International Conference on Electrical Engineering/Electronics, Computer, Telecommunications and Information Technology (ECTI-CON)","volume":"43 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129966643","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 : 2023-05-09DOI: 10.1109/ECTI-CON58255.2023.10153361
Chainarong Wanklad, P. Pairodamonchai
This paper presents common-mode (CM) and differential-mode (DM) equivalent circuits of a boost converter. It can accurately reveal the characteristics and behavior of CM and DM voltages generated by the boost converter. The analysis considers the boost converter’s circuit following the boost converter’s switching pattern. The CM and DM equations are synthesized to form the CM and DM equivalent circuits. They show the CM and DM voltage appearing at any point of the boost converter accurately and clearly. The proposed CM equivalent circuits disclose that an unbalanced boost converter and a balanced boost converter have CM voltage at the input of the boost converter differently. The CM voltage generated from the balanced boost converter has a high dv/dt value, affecting higher EMI noise. The proposed equivalent circuits can be used as a guideline for EMI analysis and designing filter circuits to resolve EMI problems for other circuits in the future.
{"title":"Common-mode and Differential-mode Equivalent Circuits for EMI Analysis in Boost Converter","authors":"Chainarong Wanklad, P. Pairodamonchai","doi":"10.1109/ECTI-CON58255.2023.10153361","DOIUrl":"https://doi.org/10.1109/ECTI-CON58255.2023.10153361","url":null,"abstract":"This paper presents common-mode (CM) and differential-mode (DM) equivalent circuits of a boost converter. It can accurately reveal the characteristics and behavior of CM and DM voltages generated by the boost converter. The analysis considers the boost converter’s circuit following the boost converter’s switching pattern. The CM and DM equations are synthesized to form the CM and DM equivalent circuits. They show the CM and DM voltage appearing at any point of the boost converter accurately and clearly. The proposed CM equivalent circuits disclose that an unbalanced boost converter and a balanced boost converter have CM voltage at the input of the boost converter differently. The CM voltage generated from the balanced boost converter has a high dv/dt value, affecting higher EMI noise. The proposed equivalent circuits can be used as a guideline for EMI analysis and designing filter circuits to resolve EMI problems for other circuits in the future.","PeriodicalId":340768,"journal":{"name":"2023 20th International Conference on Electrical Engineering/Electronics, Computer, Telecommunications and Information Technology (ECTI-CON)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131246140","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 : 2023-05-09DOI: 10.1109/ECTI-CON58255.2023.10153201
Somboon Manuch, T. Chaiyatham, N. Phansiri, Adisit Opachat, Pawarut Kongsombatsuk, W. Khanngern
Lightning locator is used to increase the efficiency of the lightning alarm system and analyze the intensity of lightning for evaluating the risk of a lightning protection system. The time difference of arrival (TDoA) technique is widely used for locating of lightning due to its high accuracy. Also, number of stations for installing the sensor is less than the other methods. However, at least 4 stations are required for calculation. Therefore, the selection of stations of sensor is very important because it affects accuracy, directly. This paper proposes the study on selection of sensors for calculation in order to locate the lightning strike. Because the TDoA is based on time difference, the window time and arrangement of sensors are considered for sensor selection. The window times is the difference of time stamp of each sensor when the lightning is detected. So that, the lowest window time of 4 sensors is applied. In addition, arrangement of sensors based on closed loop pattern can give the lowest window time. In this work, the installation of 6 networked lightning detection stations is applied for testing by 4 cases. Moreover, to ensure the proposed idea, not only difference of window time, but also the arrangement of sensors as closed loop and unclosed loop is simulated. As shown in the results, they are found that the accuracy tends to decrease when the window time increases. Also, the error highly changes when the lightning strike occurs out of loop while window time increases.
{"title":"Study on the Lightning Location System using Time Difference of Arrival Technique","authors":"Somboon Manuch, T. Chaiyatham, N. Phansiri, Adisit Opachat, Pawarut Kongsombatsuk, W. Khanngern","doi":"10.1109/ECTI-CON58255.2023.10153201","DOIUrl":"https://doi.org/10.1109/ECTI-CON58255.2023.10153201","url":null,"abstract":"Lightning locator is used to increase the efficiency of the lightning alarm system and analyze the intensity of lightning for evaluating the risk of a lightning protection system. The time difference of arrival (TDoA) technique is widely used for locating of lightning due to its high accuracy. Also, number of stations for installing the sensor is less than the other methods. However, at least 4 stations are required for calculation. Therefore, the selection of stations of sensor is very important because it affects accuracy, directly. This paper proposes the study on selection of sensors for calculation in order to locate the lightning strike. Because the TDoA is based on time difference, the window time and arrangement of sensors are considered for sensor selection. The window times is the difference of time stamp of each sensor when the lightning is detected. So that, the lowest window time of 4 sensors is applied. In addition, arrangement of sensors based on closed loop pattern can give the lowest window time. In this work, the installation of 6 networked lightning detection stations is applied for testing by 4 cases. Moreover, to ensure the proposed idea, not only difference of window time, but also the arrangement of sensors as closed loop and unclosed loop is simulated. As shown in the results, they are found that the accuracy tends to decrease when the window time increases. Also, the error highly changes when the lightning strike occurs out of loop while window time increases.","PeriodicalId":340768,"journal":{"name":"2023 20th International Conference on Electrical Engineering/Electronics, Computer, Telecommunications and Information Technology (ECTI-CON)","volume":"7 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128772854","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 : 2023-05-09DOI: 10.1109/ECTI-CON58255.2023.10153224
B. Boribun, Phaiboon Boupha, Nopparat Thammawongsa, Nuttapong Bootthanu
From 2020 – 2021, COVID-19 causes such problems in Thailand. Consequently, lockdown restriction people lead to increasing electric demand in resident areas. This work proposes a load flow analysis of the unbalanced radial distribution system influence of COVID-19 and electric vehicle (EV) charging demand. The study’s focus parameters are power quality issues such as voltage magnitude variation and power losses. The scenarios of electric vehicle charging demand also include and declare the results. Therefore, the contribution of this work is the best case study for the future operation and planning of the distribution system.
{"title":"Load Flow Analysis of Unbalanced Radial Distribution System Influence of COVID-19 Lockdowns and Electric Vehicles Charging Demand","authors":"B. Boribun, Phaiboon Boupha, Nopparat Thammawongsa, Nuttapong Bootthanu","doi":"10.1109/ECTI-CON58255.2023.10153224","DOIUrl":"https://doi.org/10.1109/ECTI-CON58255.2023.10153224","url":null,"abstract":"From 2020 – 2021, COVID-19 causes such problems in Thailand. Consequently, lockdown restriction people lead to increasing electric demand in resident areas. This work proposes a load flow analysis of the unbalanced radial distribution system influence of COVID-19 and electric vehicle (EV) charging demand. The study’s focus parameters are power quality issues such as voltage magnitude variation and power losses. The scenarios of electric vehicle charging demand also include and declare the results. Therefore, the contribution of this work is the best case study for the future operation and planning of the distribution system.","PeriodicalId":340768,"journal":{"name":"2023 20th International Conference on Electrical Engineering/Electronics, Computer, Telecommunications and Information Technology (ECTI-CON)","volume":"54 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128775224","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 : 2023-05-09DOI: 10.1109/ECTI-CON58255.2023.10153191
Thawatchai Raksachat, Rathachai Chawuthai
The purpose of this study is to develop a more effective method for categorizing Thai-language tweets related to traffic. The categorization consists of five categories. Previous studies have utilized CNN and BERT for classification, but have faced the challenge of needing balanced data for improved performance. To address this, we propose the use of BPEmb to augmentation the data and calculate cosine similarity. The subsequent step will be to create a balanced dataset to train a combination of CNN and bi-LSTM models for tweet classification. Our experiment demonstrates a significant improvement in tweet classification with a 14.3% increase in F1-score compared to the baseline method.
{"title":"Improving a text classifier using text augmentation: road traffic content from Twitter","authors":"Thawatchai Raksachat, Rathachai Chawuthai","doi":"10.1109/ECTI-CON58255.2023.10153191","DOIUrl":"https://doi.org/10.1109/ECTI-CON58255.2023.10153191","url":null,"abstract":"The purpose of this study is to develop a more effective method for categorizing Thai-language tweets related to traffic. The categorization consists of five categories. Previous studies have utilized CNN and BERT for classification, but have faced the challenge of needing balanced data for improved performance. To address this, we propose the use of BPEmb to augmentation the data and calculate cosine similarity. The subsequent step will be to create a balanced dataset to train a combination of CNN and bi-LSTM models for tweet classification. Our experiment demonstrates a significant improvement in tweet classification with a 14.3% increase in F1-score compared to the baseline method.","PeriodicalId":340768,"journal":{"name":"2023 20th International Conference on Electrical Engineering/Electronics, Computer, Telecommunications and Information Technology (ECTI-CON)","volume":"18 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125578015","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：