Pub Date : 2023-06-28DOI: 10.1109/JCSSE58229.2023.10202124
Peeradon Sukkasem, Chitsutha Soomlek
To preserve software quality and maintainability, machine learning-based code smell detection has been proposed, and the results are promising. This research proposes an enhanced version of machine learning-based code smell detection. We improve the performance of machine learning-based code smell classifiers by applying hyper-parameter optimization techniques in Particle swarm optimization and Bayesian optimization to decision tree and random forest. The models were trained and evaluated on 74 open source projects to identify god class, data class, feature envy, and long method. The experimental results confirm that the optimized machine learning classifiers c an achieve up to 99.183% and 99.155% of accuracy for both class-level and function-level code smell classification, respectively. In term of recall, the enhanced machine learning-based code smell classifiers achieved 9 9.514% when identifying data class and 98.806% for long method. The comparison results also indicated that the enhanced machine learning classifiers outperform the original versions in the code smell detection context.
{"title":"Enhanced Machine Learning-Based Code Smell Detection Through Hyper-Parameter Optimization","authors":"Peeradon Sukkasem, Chitsutha Soomlek","doi":"10.1109/JCSSE58229.2023.10202124","DOIUrl":"https://doi.org/10.1109/JCSSE58229.2023.10202124","url":null,"abstract":"To preserve software quality and maintainability, machine learning-based code smell detection has been proposed, and the results are promising. This research proposes an enhanced version of machine learning-based code smell detection. We improve the performance of machine learning-based code smell classifiers by applying hyper-parameter optimization techniques in Particle swarm optimization and Bayesian optimization to decision tree and random forest. The models were trained and evaluated on 74 open source projects to identify god class, data class, feature envy, and long method. The experimental results confirm that the optimized machine learning classifiers c an achieve up to 99.183% and 99.155% of accuracy for both class-level and function-level code smell classification, respectively. In term of recall, the enhanced machine learning-based code smell classifiers achieved 9 9.514% when identifying data class and 98.806% for long method. The comparison results also indicated that the enhanced machine learning classifiers outperform the original versions in the code smell detection context.","PeriodicalId":298838,"journal":{"name":"2023 20th International Joint Conference on Computer Science and Software Engineering (JCSSE)","volume":"113 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-06-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116305020","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}
Thai students have relatively low scores on the reading literacy assessment conducted by PISA. Various studies reported that reading skills could be improved by writing. However, essay scoring is a time-consuming task. An automated essay scoring system can support both teachers and students by reducing the teachers' workload and providing predicted scores as feedback to students. A number of recent studies have focused on automated essay scoring dataset that contains only essays written in English. Little to no research has been done on the automated essay scoring system for the Thai language. The aim of this study is to develop a Thai essay scoring system using machine learning and deep learning models that have been reported to achieve good performance. We also try to improve the performance of our models by adding essay attribute features. The models that were used in this study are logistic regression, kNN, SVM, random forest, gradient boosting, XGBoost, LSTM (bag-of-words), LSTM (w2v), BERT-based, and LSTM+CNN (BERT embedding). The models were evaluated by six metrics, including accuracy, Quadratic weighted kappa, precision, recall, and F1-score along with 10-fold cross-validation. The experimental results show that XGBoost outperforms other models considering the majority of best metric scores in each set. For deep learning models with automatically extracted features from the text, the LSTM with word2vec features model yielded better performance than other deep learning models.
{"title":"A Comparison of Machine Learning and Neural Network Algorithms for an Automated Thai Essay Scoring","authors":"Suttichai Suriyasat, Sapa Chanyachatchawan, Nuengwong Tuaycharoen","doi":"10.1109/JCSSE58229.2023.10201964","DOIUrl":"https://doi.org/10.1109/JCSSE58229.2023.10201964","url":null,"abstract":"Thai students have relatively low scores on the reading literacy assessment conducted by PISA. Various studies reported that reading skills could be improved by writing. However, essay scoring is a time-consuming task. An automated essay scoring system can support both teachers and students by reducing the teachers' workload and providing predicted scores as feedback to students. A number of recent studies have focused on automated essay scoring dataset that contains only essays written in English. Little to no research has been done on the automated essay scoring system for the Thai language. The aim of this study is to develop a Thai essay scoring system using machine learning and deep learning models that have been reported to achieve good performance. We also try to improve the performance of our models by adding essay attribute features. The models that were used in this study are logistic regression, kNN, SVM, random forest, gradient boosting, XGBoost, LSTM (bag-of-words), LSTM (w2v), BERT-based, and LSTM+CNN (BERT embedding). The models were evaluated by six metrics, including accuracy, Quadratic weighted kappa, precision, recall, and F1-score along with 10-fold cross-validation. The experimental results show that XGBoost outperforms other models considering the majority of best metric scores in each set. For deep learning models with automatically extracted features from the text, the LSTM with word2vec features model yielded better performance than other deep learning models.","PeriodicalId":298838,"journal":{"name":"2023 20th International Joint Conference on Computer Science and Software Engineering (JCSSE)","volume":"31 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-06-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127260390","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 : 2023-06-28DOI: 10.1109/JCSSE58229.2023.10201942
Nut Pinyo, Panya Lokaphadhana, Pipat Saengow, Saenyakorn Siangsanoh, Thanapoom Wonnaparhown, E. Chuangsuwanich, P. Punyabukkana, A. Suchato
Using automatic speech recognition (ASR) to transcribe videos in an online video learning platform can benefit learners in multiple ways. However, existing speech-to-text APIs can be costly to use, especially for long lecture videos commonly found in such platform. In this work, we developed a cloud-based ASR system that is cost-optimized for the workload of online learning platforms. We characterized such workload and applied a combination of techniques from system architecture, including: (1) serverless, (2) preemptible instance, and (3) batching and audio transcription optimization, including: (1) audio segmentation, (2) cost-based segment merging, and (3) locally hosted transcription model. All of which work together to provide a low transcription cost per minute of audio. We experimented and calculated the processing cost, time, and accuracy and showed that our system offers accuracy on par with existing speech-to-text services at a significantly lower cost. We have also integrated this system into an online video learning platform.
{"title":"0.01 Cent per Second: Developing a Cloud-based Cost-effective Audio Transcription System for an Online Video Learning Platform","authors":"Nut Pinyo, Panya Lokaphadhana, Pipat Saengow, Saenyakorn Siangsanoh, Thanapoom Wonnaparhown, E. Chuangsuwanich, P. Punyabukkana, A. Suchato","doi":"10.1109/JCSSE58229.2023.10201942","DOIUrl":"https://doi.org/10.1109/JCSSE58229.2023.10201942","url":null,"abstract":"Using automatic speech recognition (ASR) to transcribe videos in an online video learning platform can benefit learners in multiple ways. However, existing speech-to-text APIs can be costly to use, especially for long lecture videos commonly found in such platform. In this work, we developed a cloud-based ASR system that is cost-optimized for the workload of online learning platforms. We characterized such workload and applied a combination of techniques from system architecture, including: (1) serverless, (2) preemptible instance, and (3) batching and audio transcription optimization, including: (1) audio segmentation, (2) cost-based segment merging, and (3) locally hosted transcription model. All of which work together to provide a low transcription cost per minute of audio. We experimented and calculated the processing cost, time, and accuracy and showed that our system offers accuracy on par with existing speech-to-text services at a significantly lower cost. We have also integrated this system into an online video learning platform.","PeriodicalId":298838,"journal":{"name":"2023 20th International Joint Conference on Computer Science and Software Engineering (JCSSE)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-06-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121784782","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 : 2023-06-28DOI: 10.1109/JCSSE58229.2023.10202064
Tanakrit Jaichuen, Nanthaphop Ren, Pichai Wongapinya, S. Fugkeaw
This paper proposes the BLUR & TRACK system that anonymizes detected face based on blurring algorithm and supports efficient retrieval of the specified human face recorded in the video file. The development of our system has been driven by the privacy regulations such as GDPA and PDPA that enforce the data controllers and data processors to be aware of the high protection of personal data privacy. Most CCTVs available in the markets are not initially designed to serve the face blurring of people while the video files have been recorded. This is vulnerable to privacy breaches if those files are not strictly controlled with appropriate access control mechanisms. In this paper, our BLUR & TRACK system incorporates two major functions including blurring the human face while the video is running and the efficient tracking function that supports the face query by authorized person. To this end, we used the image frame to do face and object detection, extract the area of interest for the face and object based on region of interest (ROI). Then, we applied blurring to ROI by combining every frame that was blurred into the video. Then, they are kept in the graph database for efficient retrieval. Finally, we reported the experiment results related to the precision and recall of our proposed scheme when it was implemented with RetinaFace and YOLOv5Face models.
{"title":"BLUR & TRACK: Real-time Face Detection with Immediate Blurring and Efficient Tracking","authors":"Tanakrit Jaichuen, Nanthaphop Ren, Pichai Wongapinya, S. Fugkeaw","doi":"10.1109/JCSSE58229.2023.10202064","DOIUrl":"https://doi.org/10.1109/JCSSE58229.2023.10202064","url":null,"abstract":"This paper proposes the BLUR & TRACK system that anonymizes detected face based on blurring algorithm and supports efficient retrieval of the specified human face recorded in the video file. The development of our system has been driven by the privacy regulations such as GDPA and PDPA that enforce the data controllers and data processors to be aware of the high protection of personal data privacy. Most CCTVs available in the markets are not initially designed to serve the face blurring of people while the video files have been recorded. This is vulnerable to privacy breaches if those files are not strictly controlled with appropriate access control mechanisms. In this paper, our BLUR & TRACK system incorporates two major functions including blurring the human face while the video is running and the efficient tracking function that supports the face query by authorized person. To this end, we used the image frame to do face and object detection, extract the area of interest for the face and object based on region of interest (ROI). Then, we applied blurring to ROI by combining every frame that was blurred into the video. Then, they are kept in the graph database for efficient retrieval. Finally, we reported the experiment results related to the precision and recall of our proposed scheme when it was implemented with RetinaFace and YOLOv5Face models.","PeriodicalId":298838,"journal":{"name":"2023 20th International Joint Conference on Computer Science and Software Engineering (JCSSE)","volume":"19 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-06-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122299647","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 : 2023-06-28DOI: 10.1109/JCSSE58229.2023.10201970
Shahin Ramezany, Rachsuda Setthawong, Pisal Setthawong
Decentralized Finance (DeFi) has been on a roller coaster of swift changes for some years. In the DeFi world, the principal ingredients are smart contracts, which are often vulnerable to security issues. To work with these contracts, transaction submission is required. Arguably, advances in transaction analysis can benefit a whole Ethereum Virtual Machine (EVM) and non-EVM networks alike. Unfortunately, only a few open-source tools and commercial products exist that focus specifically on transactions. This research studies real-time event-driven EVM transaction monitoring and experiments with its effectiveness in performing the enormous task of transaction analysis and compares it with historical analysis approaches. In this study, EVM events were studied extensively, and Midnight, an open-source, full-stack proof of concept framework was used to experiment with the alternative approach. The research included a practical experiment on monitoring specific events called flash loan on the latest version of the popular DeFi protocol called AAVE.
{"title":"Midnight: An Efficient Event-driven EVM Transaction Security Monitoring Approach For Flash Loan Detection","authors":"Shahin Ramezany, Rachsuda Setthawong, Pisal Setthawong","doi":"10.1109/JCSSE58229.2023.10201970","DOIUrl":"https://doi.org/10.1109/JCSSE58229.2023.10201970","url":null,"abstract":"Decentralized Finance (DeFi) has been on a roller coaster of swift changes for some years. In the DeFi world, the principal ingredients are smart contracts, which are often vulnerable to security issues. To work with these contracts, transaction submission is required. Arguably, advances in transaction analysis can benefit a whole Ethereum Virtual Machine (EVM) and non-EVM networks alike. Unfortunately, only a few open-source tools and commercial products exist that focus specifically on transactions. This research studies real-time event-driven EVM transaction monitoring and experiments with its effectiveness in performing the enormous task of transaction analysis and compares it with historical analysis approaches. In this study, EVM events were studied extensively, and Midnight, an open-source, full-stack proof of concept framework was used to experiment with the alternative approach. The research included a practical experiment on monitoring specific events called flash loan on the latest version of the popular DeFi protocol called AAVE.","PeriodicalId":298838,"journal":{"name":"2023 20th International Joint Conference on Computer Science and Software Engineering (JCSSE)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-06-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132661567","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 : 2023-06-28DOI: 10.1109/JCSSE58229.2023.10201982
Bernardus Gery Santoso, F. Tobing, A. Kusnadi
Medical reimbursement is a reimbursement process for employee health treatment costs carried out by the company. Kompas Gramedia company has an integrated HR Portal system with SAP which was built by the reimbursement division as a forum for employees to process medical reimburse. The pandemic that has hit Indonesia has caused the Kompas Gramedia as a company to experience economic disruption. The use of SAP a company costs quite a lot every year. The company performs has an impact on the company's HR Portal system, so company perform technology displacement from SAP to ERP (Odoo). From these problems, a new medical reimbursement system was built based on ERP (Odoo) using the scrum development method. The system has been successfully built and has entered the development process by Kompas Gramedia. The ERP (Odoo) based medical reimbursement system has been evaluated using the EUCS and Likert scale, with results of 88.2% for Content, 87.7% for Accuracy, 86.6% for Format, 87.2% for Ease of Use, 87.6% for Timeliness, and 87.5% for the overall score. The reliability of the evaluation results was also measured using Cronbach's alpha and produced a value of 0.72. The conclusion can be drawn that the ERP (Odoo)-based medical reimbursement system has built is very satisfactory for the medical reimbursement process and has reliable evaluation results.
{"title":"ERP Odoo Based Medical Reimbursement System Using Scrum Method: (Study Case: Group of Retail and Publishing Kompas Gramedia)","authors":"Bernardus Gery Santoso, F. Tobing, A. Kusnadi","doi":"10.1109/JCSSE58229.2023.10201982","DOIUrl":"https://doi.org/10.1109/JCSSE58229.2023.10201982","url":null,"abstract":"Medical reimbursement is a reimbursement process for employee health treatment costs carried out by the company. Kompas Gramedia company has an integrated HR Portal system with SAP which was built by the reimbursement division as a forum for employees to process medical reimburse. The pandemic that has hit Indonesia has caused the Kompas Gramedia as a company to experience economic disruption. The use of SAP a company costs quite a lot every year. The company performs has an impact on the company's HR Portal system, so company perform technology displacement from SAP to ERP (Odoo). From these problems, a new medical reimbursement system was built based on ERP (Odoo) using the scrum development method. The system has been successfully built and has entered the development process by Kompas Gramedia. The ERP (Odoo) based medical reimbursement system has been evaluated using the EUCS and Likert scale, with results of 88.2% for Content, 87.7% for Accuracy, 86.6% for Format, 87.2% for Ease of Use, 87.6% for Timeliness, and 87.5% for the overall score. The reliability of the evaluation results was also measured using Cronbach's alpha and produced a value of 0.72. The conclusion can be drawn that the ERP (Odoo)-based medical reimbursement system has built is very satisfactory for the medical reimbursement process and has reliable evaluation results.","PeriodicalId":298838,"journal":{"name":"2023 20th International Joint Conference on Computer Science and Software Engineering (JCSSE)","volume":"32 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-06-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132175743","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 : 2023-06-28DOI: 10.1109/JCSSE58229.2023.10202090
Rohani Rohan, Pranab Roy, V. Vanijja, Suree Funilkul, Subhodeep Mukherjee, Debajyoti Pal
Metaverse is an emerging 3D digital space that augments the real world and the virtual world. It has been envisioned to be the trendsetter for future education having tremendous potential. However, the success of any new technology depends on its mass adoption and widespread societal diffusion. Therefore, it becomes important to understand the factors that can lead to its success by ensuring widespread adoption. In this work, we investigate the different psychological gratifications of the educational usage of Metaverse under the lens of the Uses and Gratification (UGT) theory. Particularly, we propose and test a research model by using the Structural Equation Modelling (SEM) approach by considering four fundamental psychological gratifications of autonomy, achievement, affiliation, and dominance, together with two additional factors of hedonic motivation and student personality. Data is collected from two Asian universities, and results show the importance of autonomy, hedonic motivation, and personality in predicting Metaverse adoption.
{"title":"What Affects the Adoption of Metaverse in Education? A SEM-based Approach","authors":"Rohani Rohan, Pranab Roy, V. Vanijja, Suree Funilkul, Subhodeep Mukherjee, Debajyoti Pal","doi":"10.1109/JCSSE58229.2023.10202090","DOIUrl":"https://doi.org/10.1109/JCSSE58229.2023.10202090","url":null,"abstract":"Metaverse is an emerging 3D digital space that augments the real world and the virtual world. It has been envisioned to be the trendsetter for future education having tremendous potential. However, the success of any new technology depends on its mass adoption and widespread societal diffusion. Therefore, it becomes important to understand the factors that can lead to its success by ensuring widespread adoption. In this work, we investigate the different psychological gratifications of the educational usage of Metaverse under the lens of the Uses and Gratification (UGT) theory. Particularly, we propose and test a research model by using the Structural Equation Modelling (SEM) approach by considering four fundamental psychological gratifications of autonomy, achievement, affiliation, and dominance, together with two additional factors of hedonic motivation and student personality. Data is collected from two Asian universities, and results show the importance of autonomy, hedonic motivation, and personality in predicting Metaverse adoption.","PeriodicalId":298838,"journal":{"name":"2023 20th International Joint Conference on Computer Science and Software Engineering (JCSSE)","volume":"116 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-06-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123687253","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 : 2023-06-28DOI: 10.1109/JCSSE58229.2023.10202122
Chanphot Wongtaweesup, Kristina Thapa, Chutiporn Anutariya, Aekavute Sujarae, James Tisyakorn
Sleep apnea is a prevalent sleep disorder that has negative consequences such as cognitive impairment, excessive sleepiness, and depression. However, it is often under-diagnosed, leading to treatment delays. Polysomnography is the standard diagnostic procedure, but it is complex and expensive, requiring specialized facilities and personnel. To address this issue, various wearable devices have been suggested for diagnosing sleep apnea in the patient's home. The purpose of this study was to compare the accuracy of Pure Health LIFE HR 2 and Wellue O2 ring with that of polysomnography for sleep apnea diagnosis. The study recruited 15 patients who were asked to wear the devices while sleeping in the Thammasat hospital sleep lab. A custom application was used to collect SpO2 (oxygen saturation) data from the smartwatch. The study found that the Pure Health LIFE HR 2 was not able to read SpO2 below 95, while the O2 ring had good accuracy with a deviation of 5% from the gold standard for SpO2 reading. The study recommends further research using machine learning and deep learning for apnea detection using O2 ring.
睡眠呼吸暂停是一种普遍存在的睡眠障碍，它会带来认知障碍、过度嗜睡和抑郁等负面影响。然而，它往往诊断不足，导致治疗延误。多导睡眠图是标准的诊断程序，但它复杂且昂贵，需要专门的设备和人员。为了解决这个问题，人们建议在患者家中使用各种可穿戴设备来诊断睡眠呼吸暂停。本研究的目的是比较Pure Health LIFE HR 2和Wellue O2环与多导睡眠图诊断睡眠呼吸暂停的准确性。这项研究招募了15名患者，他们被要求在法王医院睡眠实验室睡觉时戴上这种设备。使用自定义应用程序从智能手表收集SpO2(氧饱和度)数据。研究发现，Pure Health LIFE HR 2无法读取95以下的SpO2，而O2环具有良好的准确性，与SpO2读数的金标准偏差为5%。该研究建议进一步研究使用机器学习和深度学习来使用O2环检测呼吸暂停。
{"title":"Using Consumer-Graded Wearable Devices for Sleep Apnea Pre-Diagnosis: A Survey and Recommendations","authors":"Chanphot Wongtaweesup, Kristina Thapa, Chutiporn Anutariya, Aekavute Sujarae, James Tisyakorn","doi":"10.1109/JCSSE58229.2023.10202122","DOIUrl":"https://doi.org/10.1109/JCSSE58229.2023.10202122","url":null,"abstract":"Sleep apnea is a prevalent sleep disorder that has negative consequences such as cognitive impairment, excessive sleepiness, and depression. However, it is often under-diagnosed, leading to treatment delays. Polysomnography is the standard diagnostic procedure, but it is complex and expensive, requiring specialized facilities and personnel. To address this issue, various wearable devices have been suggested for diagnosing sleep apnea in the patient's home. The purpose of this study was to compare the accuracy of Pure Health LIFE HR 2 and Wellue O2 ring with that of polysomnography for sleep apnea diagnosis. The study recruited 15 patients who were asked to wear the devices while sleeping in the Thammasat hospital sleep lab. A custom application was used to collect SpO2 (oxygen saturation) data from the smartwatch. The study found that the Pure Health LIFE HR 2 was not able to read SpO2 below 95, while the O2 ring had good accuracy with a deviation of 5% from the gold standard for SpO2 reading. The study recommends further research using machine learning and deep learning for apnea detection using O2 ring.","PeriodicalId":298838,"journal":{"name":"2023 20th International Joint Conference on Computer Science and Software Engineering (JCSSE)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-06-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125777649","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 : 2023-06-28DOI: 10.1109/JCSSE58229.2023.10202123
Tanasit Tuangcharoentip, N. Niparnan
The authors believe that for society to continue improving, it needs to have a better waste classification system. This paper presents a computer vision model trained with a novel dataset comprising images of waste collected from Thailand, aimed at improving computer vision-based waste classification. We compare several CNN models, including VGG16, DenseNet169, ResNet-101, MobileNetV2, and InceptionV3 by using the Trashnet dataset developed by Gary Thung and our novel dataset. Our findings demonstrate that our model trained with both the Trashnet dataset and our new dataset outperforms the model trained by either dataset alone. The authors also discuss how this novel dataset can be improved to provide better results.
{"title":"Comparison of CNN Models for Urban Garbage Image Classification in Thailand","authors":"Tanasit Tuangcharoentip, N. Niparnan","doi":"10.1109/JCSSE58229.2023.10202123","DOIUrl":"https://doi.org/10.1109/JCSSE58229.2023.10202123","url":null,"abstract":"The authors believe that for society to continue improving, it needs to have a better waste classification system. This paper presents a computer vision model trained with a novel dataset comprising images of waste collected from Thailand, aimed at improving computer vision-based waste classification. We compare several CNN models, including VGG16, DenseNet169, ResNet-101, MobileNetV2, and InceptionV3 by using the Trashnet dataset developed by Gary Thung and our novel dataset. Our findings demonstrate that our model trained with both the Trashnet dataset and our new dataset outperforms the model trained by either dataset alone. The authors also discuss how this novel dataset can be improved to provide better results.","PeriodicalId":298838,"journal":{"name":"2023 20th International Joint Conference on Computer Science and Software Engineering (JCSSE)","volume":"123 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-06-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115706376","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 : 2023-06-28DOI: 10.1109/JCSSE58229.2023.10202008
Keereeluk Sirikarin, Subhorn Khonthapagdee
Water is necessary for human consumption. To ensure that water is safe, a monitoring system for water quality is required. One part of the system is to be able to predict the water quality class. Using data collected from the Pollution Control Department of Thailand from 2009 to 2021, we compared four machine learning approaches for classifying water quality classes in four main rivers in Thailand: the Ping, Wang, Yom, and Nan rivers. Random Forest, Extreme Gradient Boosting (XGBoost), Logistic Regression, and Support Vector Machine were used in this study. Moreover, synthetic minority oversampling technique (SMOTE) and Random oversampling, two strategies for dealing with imbalanced data, were also used to improve classification F1 score. This study found that XGBoost with SMOTE achieved the highest score, and BOD was the most important feature in classifying water quality.
{"title":"Machine Learning Techniques for Water Quality Classification of Thailand's Rivers","authors":"Keereeluk Sirikarin, Subhorn Khonthapagdee","doi":"10.1109/JCSSE58229.2023.10202008","DOIUrl":"https://doi.org/10.1109/JCSSE58229.2023.10202008","url":null,"abstract":"Water is necessary for human consumption. To ensure that water is safe, a monitoring system for water quality is required. One part of the system is to be able to predict the water quality class. Using data collected from the Pollution Control Department of Thailand from 2009 to 2021, we compared four machine learning approaches for classifying water quality classes in four main rivers in Thailand: the Ping, Wang, Yom, and Nan rivers. Random Forest, Extreme Gradient Boosting (XGBoost), Logistic Regression, and Support Vector Machine were used in this study. Moreover, synthetic minority oversampling technique (SMOTE) and Random oversampling, two strategies for dealing with imbalanced data, were also used to improve classification F1 score. This study found that XGBoost with SMOTE achieved the highest score, and BOD was the most important feature in classifying water quality.","PeriodicalId":298838,"journal":{"name":"2023 20th International Joint Conference on Computer Science and Software Engineering (JCSSE)","volume":"27 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-06-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114662135","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：