Pub Date : 2023-06-03DOI: 10.1080/1206212X.2023.2235458
Tanu Gupta, Ela Kumar
Medical question classification is a crucial step in developing a highly effective question-answering system for the medical field. Accurate classification of questions plays a vital role in selecting appropriate documents for answering those questions. Deep learning models, known for their ability to uncover hidden features, have gained popularity in various natural language processing (NLP) tasks. In this study, we focus on the significance of the Temporal CNN (TCN) model in extracting insightful features from biomedical questions. We propose a novel deep learning model called Bi-GRU-TCN, which combines the advantages of Bi-GRU and TCN. This model not only captures contextual features from the Bi-GRU model but also learns spatial features through TCN layers. Through a series of experiments, we evaluate our proposed approach on benchmark datasets (BioASQ 7b and 8b) using seven deep learning models, including two ensembled models. The results demonstrate that our approach shows outstanding performance in biomedical question classification, as measured by the precision, recall, F-score, and accuracy parameters.
{"title":"Fusion of Bi-GRU and temporal CNN for biomedical question classification","authors":"Tanu Gupta, Ela Kumar","doi":"10.1080/1206212X.2023.2235458","DOIUrl":"https://doi.org/10.1080/1206212X.2023.2235458","url":null,"abstract":"Medical question classification is a crucial step in developing a highly effective question-answering system for the medical field. Accurate classification of questions plays a vital role in selecting appropriate documents for answering those questions. Deep learning models, known for their ability to uncover hidden features, have gained popularity in various natural language processing (NLP) tasks. In this study, we focus on the significance of the Temporal CNN (TCN) model in extracting insightful features from biomedical questions. We propose a novel deep learning model called Bi-GRU-TCN, which combines the advantages of Bi-GRU and TCN. This model not only captures contextual features from the Bi-GRU model but also learns spatial features through TCN layers. Through a series of experiments, we evaluate our proposed approach on benchmark datasets (BioASQ 7b and 8b) using seven deep learning models, including two ensembled models. The results demonstrate that our approach shows outstanding performance in biomedical question classification, as measured by the precision, recall, F-score, and accuracy parameters.","PeriodicalId":39673,"journal":{"name":"International Journal of Computers and Applications","volume":"1 1","pages":"460 - 470"},"PeriodicalIF":0.0,"publicationDate":"2023-06-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"79797652","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 : 2023-06-03DOI: 10.1080/1206212X.2023.2223795
Bidyapati Thiyam, Shouvik Dey
The ever-growing amount of data generated by modern networks poses significant challenges for intrusion detection systems (IDS) in effectively analyzing and classifying security risks. Therefore, it is crucial to identify the most biased characteristics for building efficient and effective IDS algorithms. However, not all features are equally informative or relevant for intrusion detection. In response to these problems, this study proposes a Hybrid approach that uses traditional and advanced statistical techniques. The proposed method effectively validates the features generated from the hybrid model and set-operation theorem to provide the best optimal subset of features for IDS. Various machine learning methods are used to test the proposed model on three popular IDS datasets: NSL-KDD, UNSW NB15, and CIC-DDoS2019. The experimental findings show that the suggested hybrid technique improves IDS performance effectively and efficiently, providing a viable answer to the issues that intrusion detection systems confront.
{"title":"Statistical methods for feature selection: unlocking the key to improved accuracy","authors":"Bidyapati Thiyam, Shouvik Dey","doi":"10.1080/1206212X.2023.2223795","DOIUrl":"https://doi.org/10.1080/1206212X.2023.2223795","url":null,"abstract":"The ever-growing amount of data generated by modern networks poses significant challenges for intrusion detection systems (IDS) in effectively analyzing and classifying security risks. Therefore, it is crucial to identify the most biased characteristics for building efficient and effective IDS algorithms. However, not all features are equally informative or relevant for intrusion detection. In response to these problems, this study proposes a Hybrid approach that uses traditional and advanced statistical techniques. The proposed method effectively validates the features generated from the hybrid model and set-operation theorem to provide the best optimal subset of features for IDS. Various machine learning methods are used to test the proposed model on three popular IDS datasets: NSL-KDD, UNSW NB15, and CIC-DDoS2019. The experimental findings show that the suggested hybrid technique improves IDS performance effectively and efficiently, providing a viable answer to the issues that intrusion detection systems confront.","PeriodicalId":39673,"journal":{"name":"International Journal of Computers and Applications","volume":"93 1","pages":"433 - 443"},"PeriodicalIF":0.0,"publicationDate":"2023-06-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"77278170","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}
Image inpainting, aiming at exactly recovering missing pixels from partially observed entries, is typically an ill-posed problem. As a powerful constraint, low-rank priors have been widely applied in image inpainting to transform such problems into well-posed ones. However, the low-rank assumption of original visual data is only in an approximate mode, which in turn results in suboptimal recovery of fine-grained details, particularly when the missing rate is extremely high. Moreover, a single prior cannot faithfully capture the complex texture structure of an image. In this paper, we propose a joint usage of Smooth Tucker decomposition and Low-rank Hankel constraint (STLH) for image inpainting, which enables simultaneous capturing of the global low-rankness and local piecewise smoothness. Specifically, based on the Hankelization operation, the original image is mapped to a high-order structure for capturing more spatial and spectral information. By employing Tucker decomposition for optimizing the Hankel tensor and simultaneously applying Discrete Total Variation (DTV) to the Tucker factors, sharper edges are generated and better isotropic properties are enhanced. Moreover, to approximate the essential rank of the Tucker decomposition and avoid facing the uncertainty problem of the upper-rank limit, a reverse strategy is adopted to approximate the true rank of the Tucker decomposition. Finally, the overall image inpainting model is optimized by the well-known alternate least squares (ALS) algorithm. Extensive experiments show that the proposed method achieves state-of-the-art performance both quantitatively and qualitatively. Particularly, in the extreme case with 99% pixels missed, the results from STLH are averagely ahead of others at least 0.9dB in terms of PSNR values.
{"title":"Image inpainting via Smooth Tucker decomposition and Low-rank Hankel constraint","authors":"Jing Cai, Jiawei Jiang, Yibin Wang, Jian Zheng, Honghui Xu","doi":"10.1080/1206212X.2023.2219836","DOIUrl":"https://doi.org/10.1080/1206212X.2023.2219836","url":null,"abstract":"Image inpainting, aiming at exactly recovering missing pixels from partially observed entries, is typically an ill-posed problem. As a powerful constraint, low-rank priors have been widely applied in image inpainting to transform such problems into well-posed ones. However, the low-rank assumption of original visual data is only in an approximate mode, which in turn results in suboptimal recovery of fine-grained details, particularly when the missing rate is extremely high. Moreover, a single prior cannot faithfully capture the complex texture structure of an image. In this paper, we propose a joint usage of Smooth Tucker decomposition and Low-rank Hankel constraint (STLH) for image inpainting, which enables simultaneous capturing of the global low-rankness and local piecewise smoothness. Specifically, based on the Hankelization operation, the original image is mapped to a high-order structure for capturing more spatial and spectral information. By employing Tucker decomposition for optimizing the Hankel tensor and simultaneously applying Discrete Total Variation (DTV) to the Tucker factors, sharper edges are generated and better isotropic properties are enhanced. Moreover, to approximate the essential rank of the Tucker decomposition and avoid facing the uncertainty problem of the upper-rank limit, a reverse strategy is adopted to approximate the true rank of the Tucker decomposition. Finally, the overall image inpainting model is optimized by the well-known alternate least squares (ALS) algorithm. Extensive experiments show that the proposed method achieves state-of-the-art performance both quantitatively and qualitatively. Particularly, in the extreme case with 99% pixels missed, the results from STLH are averagely ahead of others at least 0.9dB in terms of PSNR values.","PeriodicalId":39673,"journal":{"name":"International Journal of Computers and Applications","volume":"32 1","pages":"421 - 432"},"PeriodicalIF":0.0,"publicationDate":"2023-06-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"89351765","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 : 2023-06-03DOI: 10.1080/1206212X.2023.2235750
Manabendra Nath, Pinaki S. Mitra, Deepak Kumar
Tea is one of the most valuable crops in many tea-producing countries. However, tea plants are vulnerable to various diseases, which reduce tea production. Early diagnosis of diseases is crucial to averting their detrimental effects on the growth and quality of tea. Conventional disease identification methods depend on the manual analysis of disease features by experts, which is time-consuming and resource-intensive. Moreover, published approaches based on computer vision left a broad scope for improving accuracy and reducing computational costs. This work attempts to design an automated learning-based model by leveraging the power of deep learning methods with reduced computational costs for accurately identifying tea diseases. The proposed work uses a Convolutional Neural Network architecture based on depthwise separable convolutions and residual networks integrated with a Support Vector Machine. Additionally, an attention module is added to the model for precise extraction of disease features. An image dataset is constructed comprising the images of healthy and diseased tea leaves infected with blister blight, grey blight, and red rust. The performance of the proposed model is evaluated on the self-generated tea dataset and compared with eight other state-of-the-art deep-learning models to establish its significance. The model achieves an overall accuracy of 99.28%.
{"title":"A novel residual learning-based deep learning model integrated with attention mechanism and SVM for identifying tea plant diseases","authors":"Manabendra Nath, Pinaki S. Mitra, Deepak Kumar","doi":"10.1080/1206212X.2023.2235750","DOIUrl":"https://doi.org/10.1080/1206212X.2023.2235750","url":null,"abstract":"Tea is one of the most valuable crops in many tea-producing countries. However, tea plants are vulnerable to various diseases, which reduce tea production. Early diagnosis of diseases is crucial to averting their detrimental effects on the growth and quality of tea. Conventional disease identification methods depend on the manual analysis of disease features by experts, which is time-consuming and resource-intensive. Moreover, published approaches based on computer vision left a broad scope for improving accuracy and reducing computational costs. This work attempts to design an automated learning-based model by leveraging the power of deep learning methods with reduced computational costs for accurately identifying tea diseases. The proposed work uses a Convolutional Neural Network architecture based on depthwise separable convolutions and residual networks integrated with a Support Vector Machine. Additionally, an attention module is added to the model for precise extraction of disease features. An image dataset is constructed comprising the images of healthy and diseased tea leaves infected with blister blight, grey blight, and red rust. The performance of the proposed model is evaluated on the self-generated tea dataset and compared with eight other state-of-the-art deep-learning models to establish its significance. The model achieves an overall accuracy of 99.28%.","PeriodicalId":39673,"journal":{"name":"International Journal of Computers and Applications","volume":"45 1","pages":"471 - 484"},"PeriodicalIF":0.0,"publicationDate":"2023-06-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"81298562","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-04DOI: 10.1080/1206212X.2023.2218236
Mohamed Raouf Kanfoud, Abdelkrim Bouramoul
The Web has become one of the most important data sources, and the content shared is most often multilingual, as users belong to different cultures and speak different languages. Multilingual content (document) is not suitable for many people who only need content in one language. Furthermore, dividing a multilingual document into monolingual documents helps researchers extract only the text of the desired language to use in different tasks such as training or model testing. Therefore, it is challenging to clean and divide the raw content manually. This paper presents an automatic approach to dividing a multilingual document and reassembling it into monolingual documents by examining three existing state-of-the-art tools for Language Identification (LI). We prepared different corpora with different heterogeneity characteristics for the evaluation and evaluated their code-switching pattern using three different code-switching metrics. The proposed approach reached 99% as the best accuracy result for the long segment (long text) and 90% for the mixed segment. In addition, a good correlation was found between the I-Index and accuracy with Pearson’s r = −0.998.
{"title":"Tackling the multilingual and heterogeneous documents with the pre-trained language identifiers","authors":"Mohamed Raouf Kanfoud, Abdelkrim Bouramoul","doi":"10.1080/1206212X.2023.2218236","DOIUrl":"https://doi.org/10.1080/1206212X.2023.2218236","url":null,"abstract":"The Web has become one of the most important data sources, and the content shared is most often multilingual, as users belong to different cultures and speak different languages. Multilingual content (document) is not suitable for many people who only need content in one language. Furthermore, dividing a multilingual document into monolingual documents helps researchers extract only the text of the desired language to use in different tasks such as training or model testing. Therefore, it is challenging to clean and divide the raw content manually. This paper presents an automatic approach to dividing a multilingual document and reassembling it into monolingual documents by examining three existing state-of-the-art tools for Language Identification (LI). We prepared different corpora with different heterogeneity characteristics for the evaluation and evaluated their code-switching pattern using three different code-switching metrics. The proposed approach reached 99% as the best accuracy result for the long segment (long text) and 90% for the mixed segment. In addition, a good correlation was found between the I-Index and accuracy with Pearson’s r = −0.998.","PeriodicalId":39673,"journal":{"name":"International Journal of Computers and Applications","volume":"24 2 1","pages":"391 - 402"},"PeriodicalIF":0.0,"publicationDate":"2023-05-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"88674143","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-04DOI: 10.1080/1206212X.2023.2218244
Youcef Benkhedda, F. Azouaou
The rapid expansion of social media platforms has made linking user profiles across various networks an essential aspect of maintaining a consistent identity. With 4.66 billion users reported to be in the Websphere, many are active on multiple social media platforms simultaneously. Identifying users across multiple platforms poses challenges in integrating user profiles from various sources. Different matching schemes have been suggested over the years based on different user profile features, but very little information has been uncovered about user-generated text as a unique attribute for user profile matching, which generally poses real challenges in real-world scenarios. As many users have insufficient text and the use of non-discrete text information makes the comparison operation between the two social networks of quadratic complexity. Our study examines the different existing literature schemes for matching user profile pairs based only on their generated textual content. We suggest and evaluate the effectiveness of a two stage matching approach based on Locality Sensitive Hashing clustering and nearest neighbor search. We also present other matching results of different user representations language models and matching schemes.
{"title":"Optimizing user profile matching: a text-based approach","authors":"Youcef Benkhedda, F. Azouaou","doi":"10.1080/1206212X.2023.2218244","DOIUrl":"https://doi.org/10.1080/1206212X.2023.2218244","url":null,"abstract":"The rapid expansion of social media platforms has made linking user profiles across various networks an essential aspect of maintaining a consistent identity. With 4.66 billion users reported to be in the Websphere, many are active on multiple social media platforms simultaneously. Identifying users across multiple platforms poses challenges in integrating user profiles from various sources. Different matching schemes have been suggested over the years based on different user profile features, but very little information has been uncovered about user-generated text as a unique attribute for user profile matching, which generally poses real challenges in real-world scenarios. As many users have insufficient text and the use of non-discrete text information makes the comparison operation between the two social networks of quadratic complexity. Our study examines the different existing literature schemes for matching user profile pairs based only on their generated textual content. We suggest and evaluate the effectiveness of a two stage matching approach based on Locality Sensitive Hashing clustering and nearest neighbor search. We also present other matching results of different user representations language models and matching schemes.","PeriodicalId":39673,"journal":{"name":"International Journal of Computers and Applications","volume":"1 1","pages":"403 - 412"},"PeriodicalIF":0.0,"publicationDate":"2023-05-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"73118084","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-04DOI: 10.1080/1206212X.2023.2212945
B. N. Madhukar, S. Bharathi, A. Polnaya
Numerous studies have explored different techniques for segmenting breast cancer images, in particular deep learning-based Computer-Aided Diagnosis (CAD) has recently netted attention. However, due to their down-and-out pursuance, the existing approaches like FCN (Fully Convolutional Network), PSPNet (Pyramid Scene Parsing Network), U-Net, and SegNet still required improvement for offering better semantic segmentation while identifying breast cancer. In this paper, the newly proposed breast cancer tumor segmentation method consists of four steps pre-processing, augmentation, segmenting image using multi-scale convolution and multi- attention mechanisms respectively. The proposed method utilizes the ResNet (Residual Network) backbone network with multi-scale convolution for feature map prediction. Also, the effectiveness of the multi-channel attention module with a pyramid dilated nodule is employed for semantic segmentation. Gated axial, position, and channel attention are combined to create a multi-channel attention mechanism. Additionally, War Search Optimization (WSO) algorithm is being utilized to enhance the accuracy of the segmented images. Experimentations are conducted on two datasets, viz., Breast Cancer Cell Segmentation Database and Breast Cancer Semantic Segmentation (BCSS) Database, with different existing networks. The effectiveness of the network is evaluated based on various criteria in terms of precision, accuracy, recall, (mean Intersection of Union), (Intersection of Union), etc.
许多研究探索了不同的乳腺癌图像分割技术，特别是基于深度学习的计算机辅助诊断(CAD)最近引起了人们的关注。然而，现有的FCN (Fully Convolutional Network)、PSPNet (Pyramid Scene Parsing Network)、U-Net和SegNet等方法由于其追求的不确定性，在识别乳腺癌的同时，还需要改进以提供更好的语义分割。本文提出的乳腺癌肿瘤分割方法包括预处理、增强、多尺度卷积分割和多关注分割四个步骤。该方法利用多尺度卷积的ResNet (Residual Network)骨干网进行特征映射预测。同时，利用多通道注意力模块金字塔型扩张结节的有效性进行语义分割。门控轴，位置和通道的注意相结合，以创建一个多通道的注意机制。此外，还利用战争搜索优化(WSO)算法来提高分割图像的准确性。在现有网络不同的情况下，在乳腺癌细胞分割数据库和乳腺癌语义分割数据库两个数据集上进行实验。网络的有效性是基于精度、准确度、召回率、(平均交集)、(交集)等各种标准来评估的。
{"title":"Multi-scale convolution based breast cancer image segmentation with attention mechanism in conjunction with war search optimization","authors":"B. N. Madhukar, S. Bharathi, A. Polnaya","doi":"10.1080/1206212X.2023.2212945","DOIUrl":"https://doi.org/10.1080/1206212X.2023.2212945","url":null,"abstract":"Numerous studies have explored different techniques for segmenting breast cancer images, in particular deep learning-based Computer-Aided Diagnosis (CAD) has recently netted attention. However, due to their down-and-out pursuance, the existing approaches like FCN (Fully Convolutional Network), PSPNet (Pyramid Scene Parsing Network), U-Net, and SegNet still required improvement for offering better semantic segmentation while identifying breast cancer. In this paper, the newly proposed breast cancer tumor segmentation method consists of four steps pre-processing, augmentation, segmenting image using multi-scale convolution and multi- attention mechanisms respectively. The proposed method utilizes the ResNet (Residual Network) backbone network with multi-scale convolution for feature map prediction. Also, the effectiveness of the multi-channel attention module with a pyramid dilated nodule is employed for semantic segmentation. Gated axial, position, and channel attention are combined to create a multi-channel attention mechanism. Additionally, War Search Optimization (WSO) algorithm is being utilized to enhance the accuracy of the segmented images. Experimentations are conducted on two datasets, viz., Breast Cancer Cell Segmentation Database and Breast Cancer Semantic Segmentation (BCSS) Database, with different existing networks. The effectiveness of the network is evaluated based on various criteria in terms of precision, accuracy, recall, (mean Intersection of Union), (Intersection of Union), etc.","PeriodicalId":39673,"journal":{"name":"International Journal of Computers and Applications","volume":"40 1","pages":"353 - 366"},"PeriodicalIF":0.0,"publicationDate":"2023-05-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"81942979","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-04DOI: 10.1080/1206212X.2023.2215077
nisar. ahmed, S. Gharghan, A. H. Mutlag
With growing concerns about their children’s safety and security, parents have shown an increasing interest in creating a dependable system that allows them to track and monitor their children in outdoor environments. The number of children who have gone missing, particularly in public areas, has risen, making it even more necessary to develop efficient solutions. This study focused on designing and implementing an affordable Internet of Things-based system that enables parents to track their children’s movement while they are in outdoor environments. The system described in this study relies on the use of radio frequency identification (RFID) readers installed in multiple locations to detect the presence of a child within a designated area of interest. To connect parents with this system, an Android smartphone application was developed and connected to the Thinger.io platform. The application displays the child’s location on a map using global positioning system technology and sends an alert message through the global system for mobile networks via 3G as soon as the RFID reader detects the child’s mobile tag. The system was tested within the designated area of interest to evaluate its performance. The results revealed that the RFID reader was able to detect the child’s movement within a range of approximately 4.5 meters from each RFID device and 9 meters between two RFID devices. Furthermore, the system was able to accurately determine the child’s real-time geolocation, both with and without Internet access. The system was also found to be lightweight and cost effective.
{"title":"IoT-based child tracking using RFID and GPS","authors":"nisar. ahmed, S. Gharghan, A. H. Mutlag","doi":"10.1080/1206212X.2023.2215077","DOIUrl":"https://doi.org/10.1080/1206212X.2023.2215077","url":null,"abstract":"With growing concerns about their children’s safety and security, parents have shown an increasing interest in creating a dependable system that allows them to track and monitor their children in outdoor environments. The number of children who have gone missing, particularly in public areas, has risen, making it even more necessary to develop efficient solutions. This study focused on designing and implementing an affordable Internet of Things-based system that enables parents to track their children’s movement while they are in outdoor environments. The system described in this study relies on the use of radio frequency identification (RFID) readers installed in multiple locations to detect the presence of a child within a designated area of interest. To connect parents with this system, an Android smartphone application was developed and connected to the Thinger.io platform. The application displays the child’s location on a map using global positioning system technology and sends an alert message through the global system for mobile networks via 3G as soon as the RFID reader detects the child’s mobile tag. The system was tested within the designated area of interest to evaluate its performance. The results revealed that the RFID reader was able to detect the child’s movement within a range of approximately 4.5 meters from each RFID device and 9 meters between two RFID devices. Furthermore, the system was able to accurately determine the child’s real-time geolocation, both with and without Internet access. The system was also found to be lightweight and cost effective.","PeriodicalId":39673,"journal":{"name":"International Journal of Computers and Applications","volume":"4 1","pages":"367 - 378"},"PeriodicalIF":0.0,"publicationDate":"2023-05-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"88600678","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-04DOI: 10.1080/1206212X.2023.2219117
Smail Ait El Asri, Ismail Negabi, Samir El Adib, N. Raissouni
Performing accurate extraction of buildings from remote sensing (RS) images is a crucial process with widespread applications in urban planning, disaster management, and urban monitoring. However, this task remains challenging due to the diversity and complexity of building shapes, sizes, and textures, as well as variations in lighting and weather conditions. These difficulties motivate our research to propose an improved approach for building extraction using UNet and transfer learning to address these challenges. In this work, we tested seven different backbone architectures within the UNet encoder and found that combining UNet with ResNet101 or ResNet152 yielded the best results. Based on these findings, we combined the superior performance of these base models to create a novel architecture, which achieved significant improvements over previous methods. Specifically, our method achieved a 1.33% increase in Intersection over Union (IoU) compared to the baseline UNet model. Furthermore, it demonstrated a superior performance with a 1.21% increase in IoU over UNet-ResNet101 and a 1.60% increase in IoU over UNet-ResNet152. We evaluated our proposed approach on the INRIA Aerial Image dataset and demonstrated its superiority. Our research addresses a critical need for accurate building extraction from RS images and overcomes the challenges posed by diverse building characteristics.
{"title":"Enhancing building extraction from remote sensing images through UNet and transfer learning","authors":"Smail Ait El Asri, Ismail Negabi, Samir El Adib, N. Raissouni","doi":"10.1080/1206212X.2023.2219117","DOIUrl":"https://doi.org/10.1080/1206212X.2023.2219117","url":null,"abstract":"Performing accurate extraction of buildings from remote sensing (RS) images is a crucial process with widespread applications in urban planning, disaster management, and urban monitoring. However, this task remains challenging due to the diversity and complexity of building shapes, sizes, and textures, as well as variations in lighting and weather conditions. These difficulties motivate our research to propose an improved approach for building extraction using UNet and transfer learning to address these challenges. In this work, we tested seven different backbone architectures within the UNet encoder and found that combining UNet with ResNet101 or ResNet152 yielded the best results. Based on these findings, we combined the superior performance of these base models to create a novel architecture, which achieved significant improvements over previous methods. Specifically, our method achieved a 1.33% increase in Intersection over Union (IoU) compared to the baseline UNet model. Furthermore, it demonstrated a superior performance with a 1.21% increase in IoU over UNet-ResNet101 and a 1.60% increase in IoU over UNet-ResNet152. We evaluated our proposed approach on the INRIA Aerial Image dataset and demonstrated its superiority. Our research addresses a critical need for accurate building extraction from RS images and overcomes the challenges posed by diverse building characteristics.","PeriodicalId":39673,"journal":{"name":"International Journal of Computers and Applications","volume":"442 1","pages":"413 - 419"},"PeriodicalIF":0.0,"publicationDate":"2023-05-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"86855742","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-04DOI: 10.1080/1206212X.2023.2218061
J. Liu, H. Shao, X. Deng, Y. T. Jiang
The paper discusses the high computational costs associated with convolutional neural networks (CNNs) in real-world applications due to their complex structure, primarily in hidden layers. To overcome this issue, the paper proposes a novel channel pruning technique that leverages the correlation topology of feature maps generated by each CNNs layer to construct a network with fewer nodes, reducing computational costs significantly. Redundant channels exhibit a high degree of topological similarity and tend to increase as the number of network layers rises. Removing the channel corresponding to highly correlated feature maps allows retrieval of the ‘base’ set of characteristics needed by subsequent layers. The proposed channel pruning technique provides a promising approach to reducing the computational costs of deep convolutional neural networks while maintaining high performance levels. By designing a network structure optimized for specific input data types, the method results in more efficient and effective machine learning models. The pruning operation requires fine-tuning to optimize network performance, and experiments using X-ray, chest CT, and MNIST images show that the pruned network can eliminate approximately 80% of redundant channels with minimal performance deterioration (maintaining original CNNs performance at 99.2%).
{"title":"Exploiting similarity-induced redundancies in correlation topology for channel pruning in deep convolutional neural networks","authors":"J. Liu, H. Shao, X. Deng, Y. T. Jiang","doi":"10.1080/1206212X.2023.2218061","DOIUrl":"https://doi.org/10.1080/1206212X.2023.2218061","url":null,"abstract":"The paper discusses the high computational costs associated with convolutional neural networks (CNNs) in real-world applications due to their complex structure, primarily in hidden layers. To overcome this issue, the paper proposes a novel channel pruning technique that leverages the correlation topology of feature maps generated by each CNNs layer to construct a network with fewer nodes, reducing computational costs significantly. Redundant channels exhibit a high degree of topological similarity and tend to increase as the number of network layers rises. Removing the channel corresponding to highly correlated feature maps allows retrieval of the ‘base’ set of characteristics needed by subsequent layers. The proposed channel pruning technique provides a promising approach to reducing the computational costs of deep convolutional neural networks while maintaining high performance levels. By designing a network structure optimized for specific input data types, the method results in more efficient and effective machine learning models. The pruning operation requires fine-tuning to optimize network performance, and experiments using X-ray, chest CT, and MNIST images show that the pruned network can eliminate approximately 80% of redundant channels with minimal performance deterioration (maintaining original CNNs performance at 99.2%).","PeriodicalId":39673,"journal":{"name":"International Journal of Computers and Applications","volume":"95 1","pages":"379 - 390"},"PeriodicalIF":0.0,"publicationDate":"2023-05-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"85297097","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：