Pub Date : 2023-01-01DOI: 10.12720/jait.14.5.1046-1055
Sylvester Igbo Ele, Uzoma Rita Alo, Henry Friday Nweke, Ofem Ajah Ofem
.
{"title":"Regression-Based Machine Learning Framework for Customer Churn Prediction in Telecommunication Industry","authors":"Sylvester Igbo Ele, Uzoma Rita Alo, Henry Friday Nweke, Ofem Ajah Ofem","doi":"10.12720/jait.14.5.1046-1055","DOIUrl":"https://doi.org/10.12720/jait.14.5.1046-1055","url":null,"abstract":".","PeriodicalId":36452,"journal":{"name":"Journal of Advances in Information Technology","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"136305714","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 : 2023-01-01DOI: 10.12720/jait.14.4.639-647
Lauren Al Hawi, S. Sharqawi, Q. A. Al-Haija, A. Qusef
—Cryptocurrencies like Bitcoin are one of today's financial system’s most contentious and difficult technological advances. This study aims to evaluate the performance of three different Machine Learning (ML) algorithms, namely, the Support Vector Machines (SVM), the K Nearest Neighbor (KNN), and the Light Gradient Boosted Machine (LGBM), which seeks to accurately estimate the price movement of Bitcoin, Ethereum, and Litecoin. To test these algorithms, we used an existing continuous dataset extracted from Kaggle and coinmarketcap.com. We implemented models using the Knime platform. We used auto biner for volume and market capital. Sensitivity analysis was performed to match different parameters. The F and accuracy statistics were used for the evaluation of algorithm performances. Empirical findings reveal that the KNN has the highest forecasting performance for the overall dataset in our first investigation phase. On the other hand, the SVM has the highest for forecasting Bitcoin and the LGBM for Ethereum and Litecoin in the individual dataset in the second investigation phase.
{"title":"Empirical Evaluation of Machine Learning Performance in Forecasting Cryptocurrencies","authors":"Lauren Al Hawi, S. Sharqawi, Q. A. Al-Haija, A. Qusef","doi":"10.12720/jait.14.4.639-647","DOIUrl":"https://doi.org/10.12720/jait.14.4.639-647","url":null,"abstract":"—Cryptocurrencies like Bitcoin are one of today's financial system’s most contentious and difficult technological advances. This study aims to evaluate the performance of three different Machine Learning (ML) algorithms, namely, the Support Vector Machines (SVM), the K Nearest Neighbor (KNN), and the Light Gradient Boosted Machine (LGBM), which seeks to accurately estimate the price movement of Bitcoin, Ethereum, and Litecoin. To test these algorithms, we used an existing continuous dataset extracted from Kaggle and coinmarketcap.com. We implemented models using the Knime platform. We used auto biner for volume and market capital. Sensitivity analysis was performed to match different parameters. The F and accuracy statistics were used for the evaluation of algorithm performances. Empirical findings reveal that the KNN has the highest forecasting performance for the overall dataset in our first investigation phase. On the other hand, the SVM has the highest for forecasting Bitcoin and the LGBM for Ethereum and Litecoin in the individual dataset in the second investigation phase.","PeriodicalId":36452,"journal":{"name":"Journal of Advances in Information Technology","volume":null,"pages":null},"PeriodicalIF":1.0,"publicationDate":"2023-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"66332610","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 : 2023-01-01DOI: 10.12720/jait.14.2.160-167
Amjad Rehman Khan, M. Harouni, Sepideh Sharifi, Saeed Ali Omer Bahaj, T. Saba
—Face detection and recognition in abrupt dynamic images is still challenging due to high complexity of images. To tackle this issue, we employed Gray-Level Co-occurrence Matrix (GLCM) to convert large video into smaller consequential sections containing sequence information from a series of images. GLCM is a matrix associated with the relationship between the values of adjacent pixels in an image. The proposed method is composed of two stages. First, the video is taken as input using the histogram difference method. Features are extracted using co-occurrence matrix of images, statistical methods, and the border of sudden shots extracted from the video. Second, face recognition with the Viola-Jones algorithm is performed on the sudden shots extracted in the first step. Thus, the face is extracted by video data mining in output in close shots. In this method, we compared the parameter model in three windows (3, 5 and 7) and threshold limit for detecting abrupt cuts between values (0.1, 0.5, 1.5, 1.5 and 2) for each window. The highest percentage of face detection is attained by considering the maximum percentage of abrupt cuts in the 5×5 window with a threshold value of 1.
{"title":"Face Detection in Close-up Shot Video Events Using Video Mining","authors":"Amjad Rehman Khan, M. Harouni, Sepideh Sharifi, Saeed Ali Omer Bahaj, T. Saba","doi":"10.12720/jait.14.2.160-167","DOIUrl":"https://doi.org/10.12720/jait.14.2.160-167","url":null,"abstract":"—Face detection and recognition in abrupt dynamic images is still challenging due to high complexity of images. To tackle this issue, we employed Gray-Level Co-occurrence Matrix (GLCM) to convert large video into smaller consequential sections containing sequence information from a series of images. GLCM is a matrix associated with the relationship between the values of adjacent pixels in an image. The proposed method is composed of two stages. First, the video is taken as input using the histogram difference method. Features are extracted using co-occurrence matrix of images, statistical methods, and the border of sudden shots extracted from the video. Second, face recognition with the Viola-Jones algorithm is performed on the sudden shots extracted in the first step. Thus, the face is extracted by video data mining in output in close shots. In this method, we compared the parameter model in three windows (3, 5 and 7) and threshold limit for detecting abrupt cuts between values (0.1, 0.5, 1.5, 1.5 and 2) for each window. The highest percentage of face detection is attained by considering the maximum percentage of abrupt cuts in the 5×5 window with a threshold value of 1.","PeriodicalId":36452,"journal":{"name":"Journal of Advances in Information Technology","volume":null,"pages":null},"PeriodicalIF":1.0,"publicationDate":"2023-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"66329972","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 : 2023-01-01DOI: 10.12720/jait.14.2.233-241
P. Andono, Pieter Santoso Hadi, Muljono Muljono, Catur Supriyanto
—Bahasa Indonesia is used by about 263 million people in the world but it is classified as an under-resourced language. The problem of clickbait in news analysis has gained attention in recent years. However, for Indonesian, there is still a lack of resources for clickbait tasks. Clickbait attracts the attention of readers, even though the content is not informative and misleading. The imbalance of the clickbait dataset means unequal distribution of classes within the dataset which affects the classification result. In this research, focal loss is proposed to improve classification accuracy without reducing the number of original data. Normally, clickbait data are separated into two classes, namely clickbait, and non-clickbait. However, some titles are difficult to categorize, even by humans. Therefore, this study categorizes the titles into three categories, namely clickbait, non-clickbait, and gray-clickbait. The proposed method achieves an accuracy of 93.4% in the classification of two classes, which is better than previous studies. However, the proposed method achieves an accuracy of 73.3% in the classification of three classes. Our research shows a high similarity between gray-clickbait and clickbait data, making classification more challenging. On the other hand, the use of titles on three categorizations in clickbait is not enough to provide better classification performance.
{"title":"Clickbait Detection in Indonesian News Title with Gray Unbalanced Class Based on BERT","authors":"P. Andono, Pieter Santoso Hadi, Muljono Muljono, Catur Supriyanto","doi":"10.12720/jait.14.2.233-241","DOIUrl":"https://doi.org/10.12720/jait.14.2.233-241","url":null,"abstract":"—Bahasa Indonesia is used by about 263 million people in the world but it is classified as an under-resourced language. The problem of clickbait in news analysis has gained attention in recent years. However, for Indonesian, there is still a lack of resources for clickbait tasks. Clickbait attracts the attention of readers, even though the content is not informative and misleading. The imbalance of the clickbait dataset means unequal distribution of classes within the dataset which affects the classification result. In this research, focal loss is proposed to improve classification accuracy without reducing the number of original data. Normally, clickbait data are separated into two classes, namely clickbait, and non-clickbait. However, some titles are difficult to categorize, even by humans. Therefore, this study categorizes the titles into three categories, namely clickbait, non-clickbait, and gray-clickbait. The proposed method achieves an accuracy of 93.4% in the classification of two classes, which is better than previous studies. However, the proposed method achieves an accuracy of 73.3% in the classification of three classes. Our research shows a high similarity between gray-clickbait and clickbait data, making classification more challenging. On the other hand, the use of titles on three categorizations in clickbait is not enough to provide better classification performance.","PeriodicalId":36452,"journal":{"name":"Journal of Advances in Information Technology","volume":null,"pages":null},"PeriodicalIF":1.0,"publicationDate":"2023-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"66330587","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 : 2023-01-01DOI: 10.12720/jait.14.2.350-354
R. A. C. Roque, Dhan Joseph P. Praga, G. L. Intal
—Manual processes are still evident in firms today despite the advancements in technology. Reducing manual processes can improve an organization’s competitiveness by maximizing resources and preventing disruptions. The current reimbursement process of Company ABC is a manual process that utilizes manpower, material, and financial resources. This study aims to propose an employee reimbursement system to facilitate the process using the systems analysis approach, which consists of modeling requirements, data and process modeling, object modeling, and consideration of development strategies. The JUSTINMIND software was used as the prototyping tool for the design of the user interface. The proposed process may facilitate the reimbursement process through by reducing manual workload through process automation.
{"title":"Employee Reimbursement System for a Manufacturing Company","authors":"R. A. C. Roque, Dhan Joseph P. Praga, G. L. Intal","doi":"10.12720/jait.14.2.350-354","DOIUrl":"https://doi.org/10.12720/jait.14.2.350-354","url":null,"abstract":"—Manual processes are still evident in firms today despite the advancements in technology. Reducing manual processes can improve an organization’s competitiveness by maximizing resources and preventing disruptions. The current reimbursement process of Company ABC is a manual process that utilizes manpower, material, and financial resources. This study aims to propose an employee reimbursement system to facilitate the process using the systems analysis approach, which consists of modeling requirements, data and process modeling, object modeling, and consideration of development strategies. The JUSTINMIND software was used as the prototyping tool for the design of the user interface. The proposed process may facilitate the reimbursement process through by reducing manual workload through process automation.","PeriodicalId":36452,"journal":{"name":"Journal of Advances in Information Technology","volume":null,"pages":null},"PeriodicalIF":1.0,"publicationDate":"2023-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"66330747","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 : 2023-01-01DOI: 10.12720/jait.14.3.418-425
Amjad Rehman Khan, I. Abunadi, Bayan I. Alghofaily, Haider Ali, T. Saba
I.A
一、
{"title":"Automatic Diagnosis of Rice Leaves Diseases Using Hybrid Deep Learning Model","authors":"Amjad Rehman Khan, I. Abunadi, Bayan I. Alghofaily, Haider Ali, T. Saba","doi":"10.12720/jait.14.3.418-425","DOIUrl":"https://doi.org/10.12720/jait.14.3.418-425","url":null,"abstract":"I.A","PeriodicalId":36452,"journal":{"name":"Journal of Advances in Information Technology","volume":null,"pages":null},"PeriodicalIF":1.0,"publicationDate":"2023-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"66331265","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 : 2023-01-01DOI: 10.12720/jait.14.3.550-558
Fatma Sh. El-metwally, Ali I. Eldesouky, Nahla B. Abdel-Hamid, Sally M. Elghamrawy
— A virtual assistant has a huge impact on business and an organizations development. It can be used to manage customer relations and deal with received queries, automatically reply to e-mails and phone calls.Audio signal processing has become increasingly popular since the development of virtual assistants. Deep learning and audio signal processing advancements have dramatically enhanced audio tagging. Audio Tagging (AT) is a challenge that requires eliciting descriptive labels from audio clips. This study proposes an Optimized Deep Neural Networks Audio Tagging Framework for Virtual Business Assistant to categorize and analyze audio tagging. Each input signal is used to extract the various audio tagging features. The extracted features are input into a neural network to carry out a multi-label classification for the predicted tags. Optimization techniques are used to improve the quality of the model fit for neural networks. To test the efficiency of the framework, four comparison experiments have been conducted between it and some of the others. From these results, it was concluded that this framework is better than the others in terms of efficiency. When the neural network was trained, Mel-Frequency Cepstral Coefficient (MFCC) features with Adamax achieved the best results with 93% accuracy and a 0.17% loss. When evaluating the performance of the model for seven labels, it achieved an average of precision 0.952, recall 0.952, F-score 0.951, accuracy 0.983, and an equal error rate of 0.015 in the evaluation set compared to the provided Detection and Classification of Acoustic Scenes and Events (DSCASE) baseline where he achieved and accuracy of 72.5% and
{"title":"Optimized Deep Neural Networks Audio Tagging Framework for Virtual Business Assistant","authors":"Fatma Sh. El-metwally, Ali I. Eldesouky, Nahla B. Abdel-Hamid, Sally M. Elghamrawy","doi":"10.12720/jait.14.3.550-558","DOIUrl":"https://doi.org/10.12720/jait.14.3.550-558","url":null,"abstract":"— A virtual assistant has a huge impact on business and an organizations development. It can be used to manage customer relations and deal with received queries, automatically reply to e-mails and phone calls.Audio signal processing has become increasingly popular since the development of virtual assistants. Deep learning and audio signal processing advancements have dramatically enhanced audio tagging. Audio Tagging (AT) is a challenge that requires eliciting descriptive labels from audio clips. This study proposes an Optimized Deep Neural Networks Audio Tagging Framework for Virtual Business Assistant to categorize and analyze audio tagging. Each input signal is used to extract the various audio tagging features. The extracted features are input into a neural network to carry out a multi-label classification for the predicted tags. Optimization techniques are used to improve the quality of the model fit for neural networks. To test the efficiency of the framework, four comparison experiments have been conducted between it and some of the others. From these results, it was concluded that this framework is better than the others in terms of efficiency. When the neural network was trained, Mel-Frequency Cepstral Coefficient (MFCC) features with Adamax achieved the best results with 93% accuracy and a 0.17% loss. When evaluating the performance of the model for seven labels, it achieved an average of precision 0.952, recall 0.952, F-score 0.951, accuracy 0.983, and an equal error rate of 0.015 in the evaluation set compared to the provided Detection and Classification of Acoustic Scenes and Events (DSCASE) baseline where he achieved and accuracy of 72.5% and","PeriodicalId":36452,"journal":{"name":"Journal of Advances in Information Technology","volume":null,"pages":null},"PeriodicalIF":1.0,"publicationDate":"2023-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"66332105","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 : 2023-01-01DOI: 10.12720/jait.14.4.674-684
Chakkaphat Chamnanphan, S. Vorapatratorn, Khwunta Kirimasthong, Tossapon Boongoen, Natthakan Iam-on
—The concept of a smart city and its associated services have been extensively explored in terms of innovation development and the application of technological concepts. One of the significant concerns in promoting smart living is the security of personal lives and assets, which are at risk from organized crime and acts of terrorism. A considerable amount of attention is paid to preventing bomb attacks in public places, especially the detection of an Improvised Explosive Device (IED). This research focuses on developing an analysis model that can accurately classify instances of x-ray images of baggage or objects as containing IEDs or not. The model provides an alternative to conventional techniques that fail to detect concealed or hidden devices. For this specific project, sample images are generated by experts to cover a range of cases encountered in operations during the past decade. These images are then used to develop a deep learning model, employing several data augmentation methods to overcome the issue of a limited number of training samples. As compared to a related work that exploits neural networks, the proposed model usually achieves higher accuracy rates for unseen samples, with the best accuracy rate being 0.985. Furthermore, an empirical study is conducted to determine the optimal size of the training set that exhibits good predictive performance. The study reveals that a large training set, apart from using a lot of resources, may not yield the best results as it may indicate overfitting.
{"title":"Improvised Explosive Device Detection Using CNN With X-Ray Images","authors":"Chakkaphat Chamnanphan, S. Vorapatratorn, Khwunta Kirimasthong, Tossapon Boongoen, Natthakan Iam-on","doi":"10.12720/jait.14.4.674-684","DOIUrl":"https://doi.org/10.12720/jait.14.4.674-684","url":null,"abstract":"—The concept of a smart city and its associated services have been extensively explored in terms of innovation development and the application of technological concepts. One of the significant concerns in promoting smart living is the security of personal lives and assets, which are at risk from organized crime and acts of terrorism. A considerable amount of attention is paid to preventing bomb attacks in public places, especially the detection of an Improvised Explosive Device (IED). This research focuses on developing an analysis model that can accurately classify instances of x-ray images of baggage or objects as containing IEDs or not. The model provides an alternative to conventional techniques that fail to detect concealed or hidden devices. For this specific project, sample images are generated by experts to cover a range of cases encountered in operations during the past decade. These images are then used to develop a deep learning model, employing several data augmentation methods to overcome the issue of a limited number of training samples. As compared to a related work that exploits neural networks, the proposed model usually achieves higher accuracy rates for unseen samples, with the best accuracy rate being 0.985. Furthermore, an empirical study is conducted to determine the optimal size of the training set that exhibits good predictive performance. The study reveals that a large training set, apart from using a lot of resources, may not yield the best results as it may indicate overfitting.","PeriodicalId":36452,"journal":{"name":"Journal of Advances in Information Technology","volume":null,"pages":null},"PeriodicalIF":1.0,"publicationDate":"2023-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"66332923","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 : 2023-01-01DOI: 10.12720/jait.14.4.821-829
Ruchi Sharma, P. Shrinath
—There has been an exponential increase in usage of social informatics in recent years. This makes opinion mining more complex, especially for unstructured data available online. Although a substantial amount of research has been conducted on the COVID pandemic, post-pandemic research is lacking. Our research focuses on design and implementation of opinion mining framework for unstructured data input for business intelligence dealing with post pandemic work environment in industries. In this paper, we implement opinion mining algorithm in combination with machine learning approaches providing a hybrid approach. Transformer architecture Bidirectional Encoder Representations from Transformers language model is implemented to obtain sentence level feature vector of the document corpus and t-distributed stochastic neighbor embedding is implemented for clustering experimental evaluation. In this work, performance evaluation is undertaken using the Intertopic Distance map. By applying a hybrid strategy of natural language processing and machine learning, the results of this study indicate efficient framework development and anticipated to contribute to the improvement of efficacy of opinion mining models compared to existing approaches. This research is significant and will benefit businesses in gaining valuable insights that will lead to improved decision-making and business insights.
{"title":"Improved Opinion Mining for Unstructured Data Using Machine Learning Enabling Business Intelligence","authors":"Ruchi Sharma, P. Shrinath","doi":"10.12720/jait.14.4.821-829","DOIUrl":"https://doi.org/10.12720/jait.14.4.821-829","url":null,"abstract":"—There has been an exponential increase in usage of social informatics in recent years. This makes opinion mining more complex, especially for unstructured data available online. Although a substantial amount of research has been conducted on the COVID pandemic, post-pandemic research is lacking. Our research focuses on design and implementation of opinion mining framework for unstructured data input for business intelligence dealing with post pandemic work environment in industries. In this paper, we implement opinion mining algorithm in combination with machine learning approaches providing a hybrid approach. Transformer architecture Bidirectional Encoder Representations from Transformers language model is implemented to obtain sentence level feature vector of the document corpus and t-distributed stochastic neighbor embedding is implemented for clustering experimental evaluation. In this work, performance evaluation is undertaken using the Intertopic Distance map. By applying a hybrid strategy of natural language processing and machine learning, the results of this study indicate efficient framework development and anticipated to contribute to the improvement of efficacy of opinion mining models compared to existing approaches. This research is significant and will benefit businesses in gaining valuable insights that will lead to improved decision-making and business insights.","PeriodicalId":36452,"journal":{"name":"Journal of Advances in Information Technology","volume":null,"pages":null},"PeriodicalIF":1.0,"publicationDate":"2023-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"66333622","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 : 2023-01-01DOI: 10.12720/jait.14.6.1151-1158
Abdellah El Zaar, Rachida Assawab, Ayoub Aoulalay, Nabil Benaya, Toufik Bakir, Smain Femmam, Abderrahim El Allati
—Artificial Intelligence and Deep Learning applications are well-developed as a part of human life. In the field of object recognition, Convolutional Neural Network (CNN) based methods are getting more and more important and challenging. However, existing CNN methods do not perform well on datasets that exhibit high similarities, resulting in confusion between different classes. In this study, we propose a new Deep Learning approach for recognizing date fruit categories based on the Deep Convolutional Neural Network (DCNN). The modified fine-tuning (MFTs-Net) approach can recognize with high accuracy the different date fruit categories. In order to train and to test the robustness of our proposed method, we have collected a dataset that takes into account different date fruit categories. The presented dataset is challenging as it contains classes of a unique object and presents high similarities concerning the shape, color and texture of date fruit. We show that the MFTs-Net CNN we implemented, trained and tested using the collected dataset can recognize with high accuracy the different date categories compared with state-of-the-arts works. The presented methodology works perfectly with very small datasets, which is one of the main strengths of the proposed method. Our MFTs-Net architecture performs perfectly on test data with an accuracy of 98%. 1
{"title":"MFTs-Net: A Deep Learning Approach for High Similarity Date Fruit Recognition","authors":"Abdellah El Zaar, Rachida Assawab, Ayoub Aoulalay, Nabil Benaya, Toufik Bakir, Smain Femmam, Abderrahim El Allati","doi":"10.12720/jait.14.6.1151-1158","DOIUrl":"https://doi.org/10.12720/jait.14.6.1151-1158","url":null,"abstract":"—Artificial Intelligence and Deep Learning applications are well-developed as a part of human life. In the field of object recognition, Convolutional Neural Network (CNN) based methods are getting more and more important and challenging. However, existing CNN methods do not perform well on datasets that exhibit high similarities, resulting in confusion between different classes. In this study, we propose a new Deep Learning approach for recognizing date fruit categories based on the Deep Convolutional Neural Network (DCNN). The modified fine-tuning (MFTs-Net) approach can recognize with high accuracy the different date fruit categories. In order to train and to test the robustness of our proposed method, we have collected a dataset that takes into account different date fruit categories. The presented dataset is challenging as it contains classes of a unique object and presents high similarities concerning the shape, color and texture of date fruit. We show that the MFTs-Net CNN we implemented, trained and tested using the collected dataset can recognize with high accuracy the different date categories compared with state-of-the-arts works. The presented methodology works perfectly with very small datasets, which is one of the main strengths of the proposed method. Our MFTs-Net architecture performs perfectly on test data with an accuracy of 98%. 1","PeriodicalId":36452,"journal":{"name":"Journal of Advances in Information Technology","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135610240","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：