Pub Date : 2021-12-18DOI: 10.1109/ICCIT54785.2021.9689786
Md. Kamal Ibn Shafi, Md. Rakibuz Sultan, Sheikh Md. Mushfiqur Rahman, Md. Moinul Hoque
Smart home is slowly but steadily becoming a part of our daily life in today’s world. IoT provides another dimension to it, and this should not be surprising that there are more IoT-connected devices than humans. This paper scrutinized the current state-of-the-art IoT-based smart home system and proposed a new approach using the Machine Learning(ML) technique, so that it is capable of controlling IoT devices automatically and effectively based on its prediction in real life. Synthetic data is generated, and a portion of real-time sensor data is collected to train the system controlling models. Human presence count and different environmental variables like Temperature, Humidity, and Luminosity are the features of the prediction procedure. Besides, the Controlling Levels of the models are the class attributes. The Decision Tree algorithm is implemented to classify the proposed controlling models’ data. On the other hand, Using the cross-validation technique, performance evaluation of the models is measured, illustrating the system capability.
{"title":"IoT Based Smart Home: A Machine Learning Approach","authors":"Md. Kamal Ibn Shafi, Md. Rakibuz Sultan, Sheikh Md. Mushfiqur Rahman, Md. Moinul Hoque","doi":"10.1109/ICCIT54785.2021.9689786","DOIUrl":"https://doi.org/10.1109/ICCIT54785.2021.9689786","url":null,"abstract":"Smart home is slowly but steadily becoming a part of our daily life in today’s world. IoT provides another dimension to it, and this should not be surprising that there are more IoT-connected devices than humans. This paper scrutinized the current state-of-the-art IoT-based smart home system and proposed a new approach using the Machine Learning(ML) technique, so that it is capable of controlling IoT devices automatically and effectively based on its prediction in real life. Synthetic data is generated, and a portion of real-time sensor data is collected to train the system controlling models. Human presence count and different environmental variables like Temperature, Humidity, and Luminosity are the features of the prediction procedure. Besides, the Controlling Levels of the models are the class attributes. The Decision Tree algorithm is implemented to classify the proposed controlling models’ data. On the other hand, Using the cross-validation technique, performance evaluation of the models is measured, illustrating the system capability.","PeriodicalId":166450,"journal":{"name":"2021 24th International Conference on Computer and Information Technology (ICCIT)","volume":"98 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122660615","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 : 2021-12-18DOI: 10.1109/ICCIT54785.2021.9689815
L. Paul, Sarker Saleh Ahmed Ankan, S. Shezan, Md. Zulfiker Mahmud, M. Samsuzzaman
This paper deals with a fast charging icon-shaped slotted antenna (FCISA) which is designed to operate at 2.536 GHz. This antenna covers the frequency band from 2.389 GHz to 2.6968 GHz which is applicable for 2.45 GHz Bluetooth (2.407 GHz to 2.484 GHz), 2.4 GHz Wi-Fi (2.407 GHz to 2.484 GHz), WiMAX rel 1 (2.3 GHz to 2.40 GHz) and WiMAX rel 1.5 (2.5 GHz to 2.69 GHz). The overall dimension of the antenna is $40times 38times 0.79mm^{3}$ which etched on Rogers RT 5880 ($varepsilon_{r}$=2.2, $delta$= 0.0009). The maximum gain and directivity of the antenna are 3.16 dB and 3.51 dBi respectively. The reflection coefficient of this antenna is -40.699dB with approximately unity VSWR (1.0186) which operates at a resonant frequency of 2.536 GHz. The radiation efficiency of the fast charging icon-shaped slotted antenna is always above 90%. Thus the antenna is quite appropriate for Bluetooth, Wi-Fi and WiMAX applications.
{"title":"A Fast Charging Icon-shaped Slotted Patch Antenna for Bluetooth/Wi-Fi/WiMAX Applications","authors":"L. Paul, Sarker Saleh Ahmed Ankan, S. Shezan, Md. Zulfiker Mahmud, M. Samsuzzaman","doi":"10.1109/ICCIT54785.2021.9689815","DOIUrl":"https://doi.org/10.1109/ICCIT54785.2021.9689815","url":null,"abstract":"This paper deals with a fast charging icon-shaped slotted antenna (FCISA) which is designed to operate at 2.536 GHz. This antenna covers the frequency band from 2.389 GHz to 2.6968 GHz which is applicable for 2.45 GHz Bluetooth (2.407 GHz to 2.484 GHz), 2.4 GHz Wi-Fi (2.407 GHz to 2.484 GHz), WiMAX rel 1 (2.3 GHz to 2.40 GHz) and WiMAX rel 1.5 (2.5 GHz to 2.69 GHz). The overall dimension of the antenna is $40times 38times 0.79mm^{3}$ which etched on Rogers RT 5880 ($varepsilon_{r}$=2.2, $delta$= 0.0009). The maximum gain and directivity of the antenna are 3.16 dB and 3.51 dBi respectively. The reflection coefficient of this antenna is -40.699dB with approximately unity VSWR (1.0186) which operates at a resonant frequency of 2.536 GHz. The radiation efficiency of the fast charging icon-shaped slotted antenna is always above 90%. Thus the antenna is quite appropriate for Bluetooth, Wi-Fi and WiMAX applications.","PeriodicalId":166450,"journal":{"name":"2021 24th International Conference on Computer and Information Technology (ICCIT)","volume":"53 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128066890","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 : 2021-12-18DOI: 10.1109/ICCIT54785.2021.9689808
M. A. Hossain, Md. Imrul Hasan, Md. Rashedul Islam, Nadeem Ahmed
Management of data backup and restore in case of emergency is a crucial process in every organization. This paper discusses an effective database recovery technique called Point In Time Recovery (PITR) in postgreSQL database management system. Despite emerging big data technology, relational database management system (RDBMS) is still performing the key role for storing and processing of data in most of the organizations. Almost all kinds of financial organizations like banks and mobile financial service (MFS) organizations use RDBMS as their database tool for storing their users information and all kinds of transactional information related to that organization. Nowadays those type of organizations focus on customer acquisition strategy and thus data is growing rapidly. In spite of proper system management system crash is not very uncommon while processing large volumes of data. It results loss of data and a huge financial loss for the organization. To tackle such tragedy for the business a proper data recovery system is required for every organization. Generally organizations use backup using pg_dump command and restore using pg_restore but this traditional recovery system cannot restore the data which is created or altered after the backup taken. Also this process is time inefficient because this process reconstruct the database to the state of the last dump file. Thus our research paper implements a potent process of data recovery technique in postgreSQL that can recover all data which is created or altered after the backup taken. Again this process is time efficient because it works restoring using Write Ahead log (WAL) file from the base backup.
{"title":"A Novel Recovery Process in Timelagged Server using Point in Time Recovery (PITR)","authors":"M. A. Hossain, Md. Imrul Hasan, Md. Rashedul Islam, Nadeem Ahmed","doi":"10.1109/ICCIT54785.2021.9689808","DOIUrl":"https://doi.org/10.1109/ICCIT54785.2021.9689808","url":null,"abstract":"Management of data backup and restore in case of emergency is a crucial process in every organization. This paper discusses an effective database recovery technique called Point In Time Recovery (PITR) in postgreSQL database management system. Despite emerging big data technology, relational database management system (RDBMS) is still performing the key role for storing and processing of data in most of the organizations. Almost all kinds of financial organizations like banks and mobile financial service (MFS) organizations use RDBMS as their database tool for storing their users information and all kinds of transactional information related to that organization. Nowadays those type of organizations focus on customer acquisition strategy and thus data is growing rapidly. In spite of proper system management system crash is not very uncommon while processing large volumes of data. It results loss of data and a huge financial loss for the organization. To tackle such tragedy for the business a proper data recovery system is required for every organization. Generally organizations use backup using pg_dump command and restore using pg_restore but this traditional recovery system cannot restore the data which is created or altered after the backup taken. Also this process is time inefficient because this process reconstruct the database to the state of the last dump file. Thus our research paper implements a potent process of data recovery technique in postgreSQL that can recover all data which is created or altered after the backup taken. Again this process is time efficient because it works restoring using Write Ahead log (WAL) file from the base backup.","PeriodicalId":166450,"journal":{"name":"2021 24th International Conference on Computer and Information Technology (ICCIT)","volume":"36 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122832534","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 : 2021-12-18DOI: 10.1109/ICCIT54785.2021.9689873
Md. Muhaimenur Rahman, Sarnali Basak
The pandemic of COVID-19 is spreading everywhere in the world which subsequently has led the world into the most existential health emergency, even in the second wave. Machine learning (ML) has already proved as a promising field to guide the future course of actions in healthcare as a part of combat the pandemic. In this paper, we have applied five algorithms, namely, Random Forest, Decision Tree, Ctree, Naïve Bayes, and PCA have been used to forecast the threatening death risk among the confirmed cases of Covid-19 patients. Since COVID-19 disease is more prevalent in the lungs so we’ve divided our data into two parts and applied the ML methods on it. Three different predictions have been showed by five of the ML models, where the decision tree for outcome-1, outcome-2 outperforms, and the random forest for outcome-3 performs best than the rest of all. In particular, the results show that which method works best on COVID-19 dataset as well as the prior indication of adverse health factors of the infected patient. Finally, we showed them the alive and death prediction percentage for randomly selected ten patients that demonstrate the capability of ML models. By these sorts of research, we can Figure out whether the affected people have to be taken to ICU or ought to be dealt with at home. Moreover, accuracy performance metric has been determined in two different testing set to identify the most efficient model for risk prediction.
{"title":"Risk Prediction of COVID-19 Positive Patients: How well does the machine learning tools perform?","authors":"Md. Muhaimenur Rahman, Sarnali Basak","doi":"10.1109/ICCIT54785.2021.9689873","DOIUrl":"https://doi.org/10.1109/ICCIT54785.2021.9689873","url":null,"abstract":"The pandemic of COVID-19 is spreading everywhere in the world which subsequently has led the world into the most existential health emergency, even in the second wave. Machine learning (ML) has already proved as a promising field to guide the future course of actions in healthcare as a part of combat the pandemic. In this paper, we have applied five algorithms, namely, Random Forest, Decision Tree, Ctree, Naïve Bayes, and PCA have been used to forecast the threatening death risk among the confirmed cases of Covid-19 patients. Since COVID-19 disease is more prevalent in the lungs so we’ve divided our data into two parts and applied the ML methods on it. Three different predictions have been showed by five of the ML models, where the decision tree for outcome-1, outcome-2 outperforms, and the random forest for outcome-3 performs best than the rest of all. In particular, the results show that which method works best on COVID-19 dataset as well as the prior indication of adverse health factors of the infected patient. Finally, we showed them the alive and death prediction percentage for randomly selected ten patients that demonstrate the capability of ML models. By these sorts of research, we can Figure out whether the affected people have to be taken to ICU or ought to be dealt with at home. Moreover, accuracy performance metric has been determined in two different testing set to identify the most efficient model for risk prediction.","PeriodicalId":166450,"journal":{"name":"2021 24th International Conference on Computer and Information Technology (ICCIT)","volume":"15 12","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121001744","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 : 2021-12-18DOI: 10.1109/ICCIT54785.2021.9689898
Ashif Raihan, Md. Zahed Hossain Monju, M. Hasan, Md. Tarek Habib, Md. Ismail Jabiullah, Mohammad Shorif Uddin
Automatic fish recognition is a challenging problem as far as machine vision is concerned. In any case, there is no mechanized gadget accessible that can recognize the fish and deal with an understanding in Bangladesh. This paper investigates fish recognition using multi-picture classification including deep learning procedures. For image processing and classification, TensorFlow Keras library is used in this work. The most famous image recognition deep learning model Convolutional Neural Network (CNN) is used to assess the dependability of our work. We have implemented three custom-built CNN models to see which one exhibits the best performance. To find the most effective model, the hyperparameter tuning technique is used. We have closely observed the matrix of parameters and performance to find the best model. After that model M2 is selected for real-life prediction as it has produced the highest accuracy of about 99.5%. The intended application will be helpful for the visually impaired, child, and ignorant to recognize the Bangladeshi fish.
{"title":"CNN Modeling for Recognizing Local Fish","authors":"Ashif Raihan, Md. Zahed Hossain Monju, M. Hasan, Md. Tarek Habib, Md. Ismail Jabiullah, Mohammad Shorif Uddin","doi":"10.1109/ICCIT54785.2021.9689898","DOIUrl":"https://doi.org/10.1109/ICCIT54785.2021.9689898","url":null,"abstract":"Automatic fish recognition is a challenging problem as far as machine vision is concerned. In any case, there is no mechanized gadget accessible that can recognize the fish and deal with an understanding in Bangladesh. This paper investigates fish recognition using multi-picture classification including deep learning procedures. For image processing and classification, TensorFlow Keras library is used in this work. The most famous image recognition deep learning model Convolutional Neural Network (CNN) is used to assess the dependability of our work. We have implemented three custom-built CNN models to see which one exhibits the best performance. To find the most effective model, the hyperparameter tuning technique is used. We have closely observed the matrix of parameters and performance to find the best model. After that model M2 is selected for real-life prediction as it has produced the highest accuracy of about 99.5%. The intended application will be helpful for the visually impaired, child, and ignorant to recognize the Bangladeshi fish.","PeriodicalId":166450,"journal":{"name":"2021 24th International Conference on Computer and Information Technology (ICCIT)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129034723","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 : 2021-12-18DOI: 10.1109/ICCIT54785.2021.9689864
Nadim Mahmud Dipu, Sifatul Alam Shohan, K. Salam
Optical Character Recognition (OCR) refers to the process of converting images of printed, typed, or handwritten text into machine-readable text. OCR is one of the most widely researched topics in the field of computer vision. Furthermore, highly accurate, and sophisticated Optical Character Recognition systems have been built for most of the major languages of the world such as English, French, German, Mandarin, etc. However, despite having 300 million native speakers (4.00% of the world population) and being the 5th most spoken language of the world, the Bengali language still does not have a state-of-the-art OCR system. Moreover, most of the existing systems are not able to recognize compound letters. This study strives to resolve this issue by proposing three neural network based image classification models for Bangla OCR. These models are Inception V3, VGG16, and Vision Transformer. These models have been trained on the BanglaLekha-Isolated dataset that contains 98,950 images of Bengali characters (vowels, consonants, digits, compound letters). The accuracy provided by the VGG-16, Inception V3, and Vision Transformer on the test set are 98.65%, 97.82%, and 96.88% respectively. Each of these models is much more accurate than the existing systems. Real-time implementation of these three models will be instrumental in building a state-of-the-art Bangla OCR system.
{"title":"Bangla Optical Character Recognition (OCR) Using Deep Learning Based Image Classification Algorithms","authors":"Nadim Mahmud Dipu, Sifatul Alam Shohan, K. Salam","doi":"10.1109/ICCIT54785.2021.9689864","DOIUrl":"https://doi.org/10.1109/ICCIT54785.2021.9689864","url":null,"abstract":"Optical Character Recognition (OCR) refers to the process of converting images of printed, typed, or handwritten text into machine-readable text. OCR is one of the most widely researched topics in the field of computer vision. Furthermore, highly accurate, and sophisticated Optical Character Recognition systems have been built for most of the major languages of the world such as English, French, German, Mandarin, etc. However, despite having 300 million native speakers (4.00% of the world population) and being the 5th most spoken language of the world, the Bengali language still does not have a state-of-the-art OCR system. Moreover, most of the existing systems are not able to recognize compound letters. This study strives to resolve this issue by proposing three neural network based image classification models for Bangla OCR. These models are Inception V3, VGG16, and Vision Transformer. These models have been trained on the BanglaLekha-Isolated dataset that contains 98,950 images of Bengali characters (vowels, consonants, digits, compound letters). The accuracy provided by the VGG-16, Inception V3, and Vision Transformer on the test set are 98.65%, 97.82%, and 96.88% respectively. Each of these models is much more accurate than the existing systems. Real-time implementation of these three models will be instrumental in building a state-of-the-art Bangla OCR system.","PeriodicalId":166450,"journal":{"name":"2021 24th International Conference on Computer and Information Technology (ICCIT)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129096268","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 : 2021-12-18DOI: 10.1109/ICCIT54785.2021.9689842
Md. Zesun Ahmed Mia, Md. Moinul Islam, Monjurul Haque, S. Islam, Sajidur Rahman
Nations have suffered significant economic losses as a result of non-technical electric losses resulting from power fraud. It is a criminal act of stealing electricity by applying various mechanisms that incorporate unauthorized tapping to the power line, bypassing the smart meter, etc. Electricity theft is a significant concern for not only developing countries but also developed countries as well. However, for most developing countries, the implications are catastrophic, given that their usage is always less than their demands. Electricity theft must be detected precisely and quickly in order to be mitigated. In our study, we have proposed a method of predictive ensemble machine learning techniques (IRFD) with a novel combination of feature distinction methods to detect electricity theft. In our proposed model, we have combined feature selection technique, Recursive Feature Elimination with Stratified 10-Fold cross-validation (RFECV) and Isolation Forest (IF), to identify and remove outliers along with several machine learning classifiers to forecast the theft of electricity. This study additionally enhances the management of highly imbalanced fraudulent data with Borderline-SMOTE with SVM (SVMSMOTE) and feature scaling with StandardScaler. Following the study, the Random Forest classifier observed a higher degree of accuracy (97.06%) with higher precision, recall, and F1-Score. To evaluate the efficacy of our proposed model, comparative analysis of the classification metrics is also assessed with several machine learning classifiers like Logistic Regression, Gradient Boosting, XGBoost, AdaBoost, KNN, ANN, along with Random Forest before and after fitting our proposed feature engineering techniques.
{"title":"IRFD: A Feature Engineering based Ensemble Classification for Detecting Electricity Fraud in Traditional Meters","authors":"Md. Zesun Ahmed Mia, Md. Moinul Islam, Monjurul Haque, S. Islam, Sajidur Rahman","doi":"10.1109/ICCIT54785.2021.9689842","DOIUrl":"https://doi.org/10.1109/ICCIT54785.2021.9689842","url":null,"abstract":"Nations have suffered significant economic losses as a result of non-technical electric losses resulting from power fraud. It is a criminal act of stealing electricity by applying various mechanisms that incorporate unauthorized tapping to the power line, bypassing the smart meter, etc. Electricity theft is a significant concern for not only developing countries but also developed countries as well. However, for most developing countries, the implications are catastrophic, given that their usage is always less than their demands. Electricity theft must be detected precisely and quickly in order to be mitigated. In our study, we have proposed a method of predictive ensemble machine learning techniques (IRFD) with a novel combination of feature distinction methods to detect electricity theft. In our proposed model, we have combined feature selection technique, Recursive Feature Elimination with Stratified 10-Fold cross-validation (RFECV) and Isolation Forest (IF), to identify and remove outliers along with several machine learning classifiers to forecast the theft of electricity. This study additionally enhances the management of highly imbalanced fraudulent data with Borderline-SMOTE with SVM (SVMSMOTE) and feature scaling with StandardScaler. Following the study, the Random Forest classifier observed a higher degree of accuracy (97.06%) with higher precision, recall, and F1-Score. To evaluate the efficacy of our proposed model, comparative analysis of the classification metrics is also assessed with several machine learning classifiers like Logistic Regression, Gradient Boosting, XGBoost, AdaBoost, KNN, ANN, along with Random Forest before and after fitting our proposed feature engineering techniques.","PeriodicalId":166450,"journal":{"name":"2021 24th International Conference on Computer and Information Technology (ICCIT)","volume":"41 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117247775","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 : 2021-12-18DOI: 10.1109/ICCIT54785.2021.9689843
Kamrus Salehin, M. Alam, Md. Ashifun Nabi, Fahim Ahmed, Faisal Bin Ashraf
At present, we have seen everything is getting digitized where technology almost takes full control over our life. As a result, a massive number of textual documents are generated on online platforms and news articles are no exception. People prefer to get connected with online news portals as they are updated every single hour. Newspaper articles have so many categories such as politics, sports, business, entertainment etc. Recently, we have noticed the rapid growth and increase of Bangla online news portals on the internet. It will be helpful for the online readers to get recommended the preferable news category which assists them in locating desired articles. Manually categorizing news articles takes huge time and effort. So, text categorization is necessary for the modern day, as enormous amounts of uncategorized data are an issue here. Although the study has improved in categorizing news articles greatly for languages such as English, Arabic, Chinese, Urdu, and Hindi. Among others, the Bangla language has shown little development. However, some approaches were applied to categorize Bangla news articles, using some machine learning algorithms where resources were minimum. We have applied five machine learning classifiers and two neural networks to categorize Bangla news articles where neural network LSTM performed best. To show the comparison between applied algorithms, which one is performing better, we have used four metrics that measure performance.
{"title":"A Comparative Study of Different Text Classification Approaches for Bangla News Classification","authors":"Kamrus Salehin, M. Alam, Md. Ashifun Nabi, Fahim Ahmed, Faisal Bin Ashraf","doi":"10.1109/ICCIT54785.2021.9689843","DOIUrl":"https://doi.org/10.1109/ICCIT54785.2021.9689843","url":null,"abstract":"At present, we have seen everything is getting digitized where technology almost takes full control over our life. As a result, a massive number of textual documents are generated on online platforms and news articles are no exception. People prefer to get connected with online news portals as they are updated every single hour. Newspaper articles have so many categories such as politics, sports, business, entertainment etc. Recently, we have noticed the rapid growth and increase of Bangla online news portals on the internet. It will be helpful for the online readers to get recommended the preferable news category which assists them in locating desired articles. Manually categorizing news articles takes huge time and effort. So, text categorization is necessary for the modern day, as enormous amounts of uncategorized data are an issue here. Although the study has improved in categorizing news articles greatly for languages such as English, Arabic, Chinese, Urdu, and Hindi. Among others, the Bangla language has shown little development. However, some approaches were applied to categorize Bangla news articles, using some machine learning algorithms where resources were minimum. We have applied five machine learning classifiers and two neural networks to categorize Bangla news articles where neural network LSTM performed best. To show the comparison between applied algorithms, which one is performing better, we have used four metrics that measure performance.","PeriodicalId":166450,"journal":{"name":"2021 24th International Conference on Computer and Information Technology (ICCIT)","volume":"208 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124676212","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 : 2021-12-18DOI: 10.1109/ICCIT54785.2021.9689911
A. Rahman, Lamim Ibtisam Khalid, Muntequa Imtiaz Siraji, M. M. Nishat, Fahim Faisal, Ashik Ahmed
Mental health is an essential component of human life and maintaining a healthy state is often overlooked in today’s world. While playing games online is a fantastic method to reduce stress, it imposes a negative impact on people’s mental health. For instance, anxiety disorders are a group of mental illnesses marked by intense emotions of fear and anxiety which are witnessed in online gamers to a greater extent. To aid the identification process of anxiety levels, machine learning algorithms have emerged as a handy tool. In this paper, the anxiety levels of online gamers are detected by utilizing a dataset from Kaggle by nine machine learning algorithms. The performances of the ML models have been observed through Python simulation, and comprehensive comparative analysis has been shown for both hyperparameter tuning and without hyperparameter tuning. Random search cross-validation has been brought into action for tuning the hyper parameters. In terms of several performance measures such as accuracy, precision, recall, F1-Score, and ROC-AUC, satisfactory results have been observed. It is observed that Multilayer perceptron (MLP) outperformed the other classifiers with an accuracy of 99.96%. However, Support Vector Machine (SVM) depicted promising accuracy of 99.43% whereas Gradient Boosting (GB) and XGBoost (XGB) depicted 98.54% and 98.04% accuracy respectively. Therefore, it can be concluded that with proper implementation of the ML-based diagnosis system, it is possible to detect the anxiety level of online gamers which can assist in having a deeper understanding of behaviors and impact of online gaming in daily life.
{"title":"Enhancing the Performance of Machine Learning Classifiers by Hyperparameter Optimization in Detecting Anxiety Levels of Online Gamers","authors":"A. Rahman, Lamim Ibtisam Khalid, Muntequa Imtiaz Siraji, M. M. Nishat, Fahim Faisal, Ashik Ahmed","doi":"10.1109/ICCIT54785.2021.9689911","DOIUrl":"https://doi.org/10.1109/ICCIT54785.2021.9689911","url":null,"abstract":"Mental health is an essential component of human life and maintaining a healthy state is often overlooked in today’s world. While playing games online is a fantastic method to reduce stress, it imposes a negative impact on people’s mental health. For instance, anxiety disorders are a group of mental illnesses marked by intense emotions of fear and anxiety which are witnessed in online gamers to a greater extent. To aid the identification process of anxiety levels, machine learning algorithms have emerged as a handy tool. In this paper, the anxiety levels of online gamers are detected by utilizing a dataset from Kaggle by nine machine learning algorithms. The performances of the ML models have been observed through Python simulation, and comprehensive comparative analysis has been shown for both hyperparameter tuning and without hyperparameter tuning. Random search cross-validation has been brought into action for tuning the hyper parameters. In terms of several performance measures such as accuracy, precision, recall, F1-Score, and ROC-AUC, satisfactory results have been observed. It is observed that Multilayer perceptron (MLP) outperformed the other classifiers with an accuracy of 99.96%. However, Support Vector Machine (SVM) depicted promising accuracy of 99.43% whereas Gradient Boosting (GB) and XGBoost (XGB) depicted 98.54% and 98.04% accuracy respectively. Therefore, it can be concluded that with proper implementation of the ML-based diagnosis system, it is possible to detect the anxiety level of online gamers which can assist in having a deeper understanding of behaviors and impact of online gaming in daily life.","PeriodicalId":166450,"journal":{"name":"2021 24th International Conference on Computer and Information Technology (ICCIT)","volume":"238 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132759301","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 : 2021-12-18DOI: 10.1109/ICCIT54785.2021.9689882
Sadia Zaman Mishu, Md. Al Mamun, Md. Ali Hossain
Dimension reduction from higher dimensional hyperspectral image (HSI) data cube has grown into a significant area of research for efficient classification of ground objects. The HSI data cube is a set of numerous highly correlated narrow spectral bands. For effective classification of hyperspectral image, dimension reduction strategies are performed using feature extraction and/or feature selection methods. Standard unsupervised feature extraction method Principal Component Analysis (PCA) has been used frequently for band reduction. But PCA suffers from limitation such as failure of extracting inherent structure of HSI data because of its global variance dependency. Folded-Principal Component Analysis (FPCA), an improvement of PCA, overcomes this problem by considering both the global and local structures of HSI with less computation and memory requirements. In this paper, a hybrid approach is proposed where FPCA is applied to produce new features from the original spectra bands. Then feature selection is performed on the extracted features using normalized Mutual Information (nMI) to select the relevant features. Finally, Kernel-Support Vector Machine (K-SVM) is applied to estimate the classification accuracy of the reduced data cube. The proposed method (FPCA-nMI) is assessed on a real mixed agricultural dataset and achieved the highest classification accuracy of 97.92%, outperforming the baseline approaches.
{"title":"Folded-PCA Based Hybrid Dimension Reduction for Effective Classification of Hyperspectral Image","authors":"Sadia Zaman Mishu, Md. Al Mamun, Md. Ali Hossain","doi":"10.1109/ICCIT54785.2021.9689882","DOIUrl":"https://doi.org/10.1109/ICCIT54785.2021.9689882","url":null,"abstract":"Dimension reduction from higher dimensional hyperspectral image (HSI) data cube has grown into a significant area of research for efficient classification of ground objects. The HSI data cube is a set of numerous highly correlated narrow spectral bands. For effective classification of hyperspectral image, dimension reduction strategies are performed using feature extraction and/or feature selection methods. Standard unsupervised feature extraction method Principal Component Analysis (PCA) has been used frequently for band reduction. But PCA suffers from limitation such as failure of extracting inherent structure of HSI data because of its global variance dependency. Folded-Principal Component Analysis (FPCA), an improvement of PCA, overcomes this problem by considering both the global and local structures of HSI with less computation and memory requirements. In this paper, a hybrid approach is proposed where FPCA is applied to produce new features from the original spectra bands. Then feature selection is performed on the extracted features using normalized Mutual Information (nMI) to select the relevant features. Finally, Kernel-Support Vector Machine (K-SVM) is applied to estimate the classification accuracy of the reduced data cube. The proposed method (FPCA-nMI) is assessed on a real mixed agricultural dataset and achieved the highest classification accuracy of 97.92%, outperforming the baseline approaches.","PeriodicalId":166450,"journal":{"name":"2021 24th International Conference on Computer and Information Technology (ICCIT)","volume":"9 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133694314","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：