Computer vision and deep learning techniques are the most emerging technologies in this era. Both of these can greatly raise the rate at which defects on metal surfaces are identified while performing industrial quality checks. The identification of faults over metal surfaces can be viewed as a significant challenge since they are easily impacted by ambient factors including illumination and light reflections. This paper proposes novel metal surface defect detection network called as YOLOv-5s-FRN in response to the problems of ineffective detection brought by the conventional manual inspection system. The proposed system is developed through the integration of a novel architectural module called as Feature Recalibration Network (FRN) to the You Only Look Once-version-5 small network )YOLOv-5s(. In order to extract the global feature information from the provided image, FRN is able to evaluate the interdependencies between the channels. This improves the feature discrimination capability and prediction accuracy of the defect detection system. The incorporation of FRN structure makes YOLOv-5s architecture to selectively enhance the necessary features and discard the unwanted ones. Therefore, the proposed novel method will efficiently detect and classify the metal surface defects such as crazing, patches, inclusions, scratches, pitted surfaces and rolled in scale. North Eastern University Surface Defect Database (NEU-DET) has been used to train and test the proposed architectural model. The suggested system has been compared with alternative models based on several performance matrices such as precision, recall and Mean Average Precision (mAP). It is observed that the proposed YOLOv-5s-FRN architecture provides significant performance improvement than state-of-the-art methods. The proposed system has been provided satisfactory results by means of improvement in mAP and time consumption. The proposed model has delivered value of mAP_0.5 as 98.05% and that of mAP_0.5:0.95 as 89.03%.
{"title":"Deep Learning Based Feature Discriminability Boosted Concurrent Metal Surface Defect Detection System Using YOLOv-5s-FRN","authors":"Reshma P. Vengaloor, Roopa Muralidhar","doi":"10.34028//iajit/21/1/9","DOIUrl":"https://doi.org/10.34028//iajit/21/1/9","url":null,"abstract":"Computer vision and deep learning techniques are the most emerging technologies in this era. Both of these can greatly raise the rate at which defects on metal surfaces are identified while performing industrial quality checks. The identification of faults over metal surfaces can be viewed as a significant challenge since they are easily impacted by ambient factors including illumination and light reflections. This paper proposes novel metal surface defect detection network called as YOLOv-5s-FRN in response to the problems of ineffective detection brought by the conventional manual inspection system. The proposed system is developed through the integration of a novel architectural module called as Feature Recalibration Network (FRN) to the You Only Look Once-version-5 small network )YOLOv-5s(. In order to extract the global feature information from the provided image, FRN is able to evaluate the interdependencies between the channels. This improves the feature discrimination capability and prediction accuracy of the defect detection system. The incorporation of FRN structure makes YOLOv-5s architecture to selectively enhance the necessary features and discard the unwanted ones. Therefore, the proposed novel method will efficiently detect and classify the metal surface defects such as crazing, patches, inclusions, scratches, pitted surfaces and rolled in scale. North Eastern University Surface Defect Database (NEU-DET) has been used to train and test the proposed architectural model. The suggested system has been compared with alternative models based on several performance matrices such as precision, recall and Mean Average Precision (mAP). It is observed that the proposed YOLOv-5s-FRN architecture provides significant performance improvement than state-of-the-art methods. The proposed system has been provided satisfactory results by means of improvement in mAP and time consumption. The proposed model has delivered value of mAP_0.5 as 98.05% and that of mAP_0.5:0.95 as 89.03%.","PeriodicalId":161392,"journal":{"name":"The International Arab Journal of Information Technology","volume":"10 2","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139129337","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}
Event relation extraction is a key aspect in the fields of event evolutionary graph construction, knowledge question and answer, and intelligence analysis, etc. Currently, supervised learning methods that rely on large amounts of labeled data are mostly used; however, the size of existing event relation datasets is small and cannot provide sufficient training data for the models. To alleviate this challenging research question, this study proposes a novel data augmentation model, called Event Relation Data Augmentation based on relationship Prediction (ERDAP), that allows both semantic and structural features to be taken into account without changing the semantic relation label compatibility, uses event relation graph convolutional neural networks to predict event relations, and expands the generated high-quality event relation triples as new training data for the event relation texts. Experimental evaluation using event causality extraction method on Chinese emergent event dataset shows that our model significantly outperforms existing text augmentation methods and achieves desirable performance, which provides a new idea for event relation data augmentation
{"title":"ERDAP: A Novel Method of Event Relation Data Augmentation Based on Relation Prediction","authors":"Ruijuan Hu, Yue Chen, Haiyan Liu","doi":"10.34028/iajit/21/1/6","DOIUrl":"https://doi.org/10.34028/iajit/21/1/6","url":null,"abstract":"Event relation extraction is a key aspect in the fields of event evolutionary graph construction, knowledge question and answer, and intelligence analysis, etc. Currently, supervised learning methods that rely on large amounts of labeled data are mostly used; however, the size of existing event relation datasets is small and cannot provide sufficient training data for the models. To alleviate this challenging research question, this study proposes a novel data augmentation model, called Event Relation Data Augmentation based on relationship Prediction (ERDAP), that allows both semantic and structural features to be taken into account without changing the semantic relation label compatibility, uses event relation graph convolutional neural networks to predict event relations, and expands the generated high-quality event relation triples as new training data for the event relation texts. Experimental evaluation using event causality extraction method on Chinese emergent event dataset shows that our model significantly outperforms existing text augmentation methods and achieves desirable performance, which provides a new idea for event relation data augmentation","PeriodicalId":161392,"journal":{"name":"The International Arab Journal of Information Technology","volume":"9 14","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139129675","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}
Cloud application practitioners are building large-scale enterprise applications as microservices, to leverage scalability, performance, and availability. Microservices architecture allows a large monolithic application to be split into small, loosely coupled services. A service communicates with other services using lightweight protocols such as RESTful APIs. Extracting microservices from the monolith is a challenging task and is mostly performed manually by system architects based on their skills. This extraction involves both: 1) Partitioning of business logic, 2) Partitioning of database. For partitioning of business logic, the existing research studies focus on decomposition by considering the dependencies in the application at the class-level. However, with the passage of time, monolith application classes outgrow their size defying the Single Responsibility Principle (SRP). So, there is a need to consider the code within the classes when identifying microservices. Current studies also lack the partitioning of database and ignore the mapping of Database Entities (DE) to the microservices. In this paper, we present a Comprehensive Microservice Extraction Approach (CMEA) that considers: 1) Both classes and their methods to define and refine microservices, 2) Associate the DE to microservices using newly devised eight guiding rules handling ownership conflicts. This approach has been applied to three benchmark web applications implemented in Java and one in-house application implemented in both Java and Python. Our results demonstrate better or similar software quality attributes in comparison to the existing related studies. CMEA improves software quality attributes by 22%. System architects can easily identify microservices along with their DE using our approach. The CMEA is generic and language-independent so it can be used for any application
{"title":"A Comprehensive Microservice Extraction Approach Integrating Business Functions and Database Entities","authors":"Deepali Bajaj, Anita Goel, Suresh Gupta","doi":"10.34028/iajit/21/1/3","DOIUrl":"https://doi.org/10.34028/iajit/21/1/3","url":null,"abstract":"Cloud application practitioners are building large-scale enterprise applications as microservices, to leverage scalability, performance, and availability. Microservices architecture allows a large monolithic application to be split into small, loosely coupled services. A service communicates with other services using lightweight protocols such as RESTful APIs. Extracting microservices from the monolith is a challenging task and is mostly performed manually by system architects based on their skills. This extraction involves both: 1) Partitioning of business logic, 2) Partitioning of database. For partitioning of business logic, the existing research studies focus on decomposition by considering the dependencies in the application at the class-level. However, with the passage of time, monolith application classes outgrow their size defying the Single Responsibility Principle (SRP). So, there is a need to consider the code within the classes when identifying microservices. Current studies also lack the partitioning of database and ignore the mapping of Database Entities (DE) to the microservices. In this paper, we present a Comprehensive Microservice Extraction Approach (CMEA) that considers: 1) Both classes and their methods to define and refine microservices, 2) Associate the DE to microservices using newly devised eight guiding rules handling ownership conflicts. This approach has been applied to three benchmark web applications implemented in Java and one in-house application implemented in both Java and Python. Our results demonstrate better or similar software quality attributes in comparison to the existing related studies. CMEA improves software quality attributes by 22%. System architects can easily identify microservices along with their DE using our approach. The CMEA is generic and language-independent so it can be used for any application","PeriodicalId":161392,"journal":{"name":"The International Arab Journal of Information Technology","volume":"75 4","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139128150","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}
In medical imaging, the effective detection and classification of Breast Cancer (BC) is a current research important task because of the still existing difficulty to distinguish abnormalities from normal breast tissues due to their subtle appearance and ambiguous margins and distinguish abnormalities from the normal breast. Moreover, BC detection based on an automated detection model is needed, because manual diagnosis faces problems due to cost and shortage of skilled manpower, and also takes a very long time. Using deep learning and ensemble feature selection techniques, in this paper, a novel framework is introduced for classifying BC from histopathology images. The five primary steps of the suggested framework are as follows: 1) to make the largest original dataset and then deep learning model with data augmentation to improve the learning. 2) The best features are selected by an Ensemble Filter Feature selection Method (EFFM) which combines the best feature subsets to produce the final feature subsets. 3) Then the pruned Convolution Neural Network (CNN) model is utilized to extract the optimal features. 4) Finally, the classification is done through the Triplet Attention based Efficient Network (TAENet) classifier. The suggested model produces a 98% accuracy rate after being trained and tested on two different histopathology imaging datasets including images from four different data cohorts. Subsequently, the suggested strategy outperforms the conventional ones since the ensemble filter habitually acquires the best features, and experimental results demonstrate the importance of the proposed approach
在医学影像领域，乳腺癌（BC）的有效检测和分类是当前研究的一项重要任务，因为异常乳腺组织的外观细微，边缘模糊不清，将异常乳腺组织与正常乳腺组织区分开来仍然存在困难。此外，由于人工诊断面临成本高、技术人才短缺等问题，而且耗时很长，因此需要基于自动检测模型的乳腺癌检测。本文利用深度学习和集合特征选择技术，介绍了一种从组织病理学图像中对 BC 进行分类的新型框架。建议框架的五个主要步骤如下：1) 制作最大的原始数据集，然后通过数据增强的深度学习模型来提高学习效果。2）通过集合过滤特征选择法（EFFM）选出最佳特征，该方法将最佳特征子集结合起来，产生最终特征子集。3) 然后利用剪枝卷积神经网络（CNN）模型提取最佳特征。4) 最后，通过基于三重注意的高效网络（TAENet）分类器进行分类。建议的模型在两个不同的组织病理学成像数据集（包括来自四个不同数据队列的图像）上经过训练和测试后，准确率达到 98%。随后，由于集合滤波器习惯性地获取最佳特征，因此建议的策略优于传统策略，实验结果证明了建议方法的重要性。
{"title":"Classification of Breast Cancer using Ensemble Filter Feature Selection with Triplet Attention Based Efficient Net Classifier","authors":"Madhukar Bangalore Nagaraj, Bharathi Shivanandamurthy Hiremath, Ashwin Matta Polnaya","doi":"10.34028/iajit/21/1/2","DOIUrl":"https://doi.org/10.34028/iajit/21/1/2","url":null,"abstract":"In medical imaging, the effective detection and classification of Breast Cancer (BC) is a current research important task because of the still existing difficulty to distinguish abnormalities from normal breast tissues due to their subtle appearance and ambiguous margins and distinguish abnormalities from the normal breast. Moreover, BC detection based on an automated detection model is needed, because manual diagnosis faces problems due to cost and shortage of skilled manpower, and also takes a very long time. Using deep learning and ensemble feature selection techniques, in this paper, a novel framework is introduced for classifying BC from histopathology images. The five primary steps of the suggested framework are as follows: 1) to make the largest original dataset and then deep learning model with data augmentation to improve the learning. 2) The best features are selected by an Ensemble Filter Feature selection Method (EFFM) which combines the best feature subsets to produce the final feature subsets. 3) Then the pruned Convolution Neural Network (CNN) model is utilized to extract the optimal features. 4) Finally, the classification is done through the Triplet Attention based Efficient Network (TAENet) classifier. The suggested model produces a 98% accuracy rate after being trained and tested on two different histopathology imaging datasets including images from four different data cohorts. Subsequently, the suggested strategy outperforms the conventional ones since the ensemble filter habitually acquires the best features, and experimental results demonstrate the importance of the proposed approach","PeriodicalId":161392,"journal":{"name":"The International Arab Journal of Information Technology","volume":"46 5","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139126993","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}
Mustafa Al-Fayoumi, Q. Abu Al-haija, Rakan Armoush, Christine Amareen
With the increasing number of malicious PDF files used for cyberattacks, it is essential to develop efficient and accurate classifiers to detect and prevent these threats. Machine Learning (ML) models have successfully detected malicious PDF files. This paper presents XAI-PDF, an efficient system for malicious PDF detection designed to enhance accuracy and minimize decision-making time on a modern dataset, the Evasive-PDFMal2022 dataset. The proposed method optimizes malicious PDF classifier performance by employing feature engineering guided by Shapley Additive Explanations (SHAP). Particularly, the model development approach comprises four phases: data preparation, model building, explainability of the models, and derived features. Utilizing the interpretability of SHAP values, crucial features are identified, and new ones are generated, resulting in an improved classification model that showcases the effectiveness of interpretable AI techniques in enhancing model performance. Various interpretable ML models were implemented, with the Lightweight Gradient Boosting Machine (LGBM) outperforming other classifiers. The Explainable Artificial Intelligence (XAI) global surrogate model generated explanations for LGBM predictions. Experimental comparisons of XAI-PDF with baseline methods revealed its superiority in achieving higher accuracy, precision, and F1-scores with minimal False Positive (FP) and False Negative (FN) rates (99.9%, 100%, 99.89%,0.000, and 0.002, respectively). Additionally, XAI-PDF requires only 1.36 milliseconds per record for predictions, demonstrating increased resilience in detecting evasive malicious PDF files compared to state-of-the-art methods
随着用于网络攻击的恶意 PDF 文件数量不断增加，开发高效、准确的分类器来检测和预防这些威胁至关重要。机器学习（ML）模型已成功检测出恶意 PDF 文件。本文介绍了 XAI-PDF，这是一种高效的恶意 PDF 检测系统，旨在提高准确性并最大限度地减少现代数据集（Evasive-PDFMal2022 数据集）上的决策时间。所提出的方法通过采用以 Shapley Additive Explanations (SHAP) 为指导的特征工程来优化恶意 PDF 分类器的性能。特别是，模型开发方法包括四个阶段：数据准备、模型构建、模型的可解释性和衍生特征。利用 SHAP 值的可解释性，确定关键特征并生成新特征，从而改进分类模型，展示可解释人工智能技术在提高模型性能方面的有效性。我们实施了各种可解释的 ML 模型，其中轻量级梯度提升机（LGBM）的表现优于其他分类器。可解释人工智能（XAI）全局代理模型为 LGBM 预测生成了解释。XAI-PDF 与基线方法的实验比较显示，XAI-PDF 在实现更高的准确度、精确度和 F1 分数以及最低的假阳性（FP）和假阴性（FN）率（分别为 99.9%、100%、99.89%、0.000 和 0.002）方面更具优势。此外，XAI-PDF 的每条预测记录仅需 1.36 毫秒，这表明与最先进的方法相比，XAI-PDF 在检测躲避性恶意 PDF 文件方面具有更强的适应性。
{"title":"XAI-PDF: A Robust Framework for Malicious PDF Detection Leveraging SHAP-Based Feature Engineering","authors":"Mustafa Al-Fayoumi, Q. Abu Al-haija, Rakan Armoush, Christine Amareen","doi":"10.34028/iajit/21/1/12","DOIUrl":"https://doi.org/10.34028/iajit/21/1/12","url":null,"abstract":"With the increasing number of malicious PDF files used for cyberattacks, it is essential to develop efficient and accurate classifiers to detect and prevent these threats. Machine Learning (ML) models have successfully detected malicious PDF files. This paper presents XAI-PDF, an efficient system for malicious PDF detection designed to enhance accuracy and minimize decision-making time on a modern dataset, the Evasive-PDFMal2022 dataset. The proposed method optimizes malicious PDF classifier performance by employing feature engineering guided by Shapley Additive Explanations (SHAP). Particularly, the model development approach comprises four phases: data preparation, model building, explainability of the models, and derived features. Utilizing the interpretability of SHAP values, crucial features are identified, and new ones are generated, resulting in an improved classification model that showcases the effectiveness of interpretable AI techniques in enhancing model performance. Various interpretable ML models were implemented, with the Lightweight Gradient Boosting Machine (LGBM) outperforming other classifiers. The Explainable Artificial Intelligence (XAI) global surrogate model generated explanations for LGBM predictions. Experimental comparisons of XAI-PDF with baseline methods revealed its superiority in achieving higher accuracy, precision, and F1-scores with minimal False Positive (FP) and False Negative (FN) rates (99.9%, 100%, 99.89%,0.000, and 0.002, respectively). Additionally, XAI-PDF requires only 1.36 milliseconds per record for predictions, demonstrating increased resilience in detecting evasive malicious PDF files compared to state-of-the-art methods","PeriodicalId":161392,"journal":{"name":"The International Arab Journal of Information Technology","volume":"43 8","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139126539","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This study is centered on a significant scientific contribution within the realm of image cryptography. The chosen approach involves employing the bidimensional Arnold Cat Map transformation to reposition and modify pixel locations, guided by parameters derived from the original image. The construction of the multiplicative group Z/nZ, comprising equivalence classes modulo n, relies on a hyper-chaotic sequence derived from the 2D sinusoidal logistic modulation map. The correlation between this sequence and the preceding step yields an unpredictable blurring pattern, effectively altering the statistical properties of resulting matrices and distributing the influence of individual bits across the entire encrypted image. For each pixel, the encryption process entails an XOR operation with the Z/nZ group, followed by a right shift based on the three Least Significant Bits (LSB) of the preceding pixel. This meticulous procedure is iterated for every pixel, leaving no trace of similarity or association with the original plaintext image, effectively rendering it blurred and indecipherable. To gauge the efficacy of our algorithm, we subjected it to thorough evaluation utilizing diverse criteria, including histogram analysis, which unveiled a nearly uniform pattern in the encrypted images. Entropy values were found to be close to 8, while the correlation analysis exhibited a pronounced proximity to 0. Moreover, we subjected our approach to differential attacks, and the calculated values of the Number of Changing Pixel Rate (NPCR > 99.6) and the Unified Averaged Changed Intensity (UACI > 33.2) corroborated the strength and resilience of our methodology. In addition, to establish its comparative standing, we undertook a comprehensive assessment, meticulously comparing our method to various existing approaches from the literature, including those proposed by Hua, Es-sabry, and Faragallah. This systematic process accentuated the high level of responsiveness and sensitivity inherent in our approach, thus underscoring its innovative and promising nature
{"title":"The Coupling of a Multiplicative Group and the Theory of Chaos in the Encryptions of Images","authors":"F. ElAzzaby, N. Elakkad, Khalid Sabour","doi":"10.34028/iajit/21/1/1","DOIUrl":"https://doi.org/10.34028/iajit/21/1/1","url":null,"abstract":"This study is centered on a significant scientific contribution within the realm of image cryptography. The chosen approach involves employing the bidimensional Arnold Cat Map transformation to reposition and modify pixel locations, guided by parameters derived from the original image. The construction of the multiplicative group Z/nZ, comprising equivalence classes modulo n, relies on a hyper-chaotic sequence derived from the 2D sinusoidal logistic modulation map. The correlation between this sequence and the preceding step yields an unpredictable blurring pattern, effectively altering the statistical properties of resulting matrices and distributing the influence of individual bits across the entire encrypted image. For each pixel, the encryption process entails an XOR operation with the Z/nZ group, followed by a right shift based on the three Least Significant Bits (LSB) of the preceding pixel. This meticulous procedure is iterated for every pixel, leaving no trace of similarity or association with the original plaintext image, effectively rendering it blurred and indecipherable. To gauge the efficacy of our algorithm, we subjected it to thorough evaluation utilizing diverse criteria, including histogram analysis, which unveiled a nearly uniform pattern in the encrypted images. Entropy values were found to be close to 8, while the correlation analysis exhibited a pronounced proximity to 0. Moreover, we subjected our approach to differential attacks, and the calculated values of the Number of Changing Pixel Rate (NPCR > 99.6) and the Unified Averaged Changed Intensity (UACI > 33.2) corroborated the strength and resilience of our methodology. In addition, to establish its comparative standing, we undertook a comprehensive assessment, meticulously comparing our method to various existing approaches from the literature, including those proposed by Hua, Es-sabry, and Faragallah. This systematic process accentuated the high level of responsiveness and sensitivity inherent in our approach, thus underscoring its innovative and promising nature","PeriodicalId":161392,"journal":{"name":"The International Arab Journal of Information Technology","volume":"44 7","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139127951","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}
Semi-supervised learning is a powerful paradigm for excavating latent structures of between labeled and unlabeled samples under the view of models constructing. Currently, graph-based models solve the approximate matrix that directly represent distributions of samples by the spatial metric. The crux lies in optimizing connections of samples, which is achieved by subjecting to must-links or cannot-links. Unfortunately, to find links are rather difficult for semi-supervised clustering with very few labeled samples, therefore, significantly impairs the robustness and accuracy in such scenario. To address this problem, we propose the Cohesive Pair-wises Constrained deep Embedding model (CPCE) to obtain an optimal embedding for representing the category distribution of samples and avoid the failed graph-structure of the global samples. CPCE designs the deep network framework based on CNN-Autoencoder by minimizing reconstruct errors of samples, and build up constrains both of the sample distribution for within-class and the category distribution for intra-class to optimal the latent embedding. Then, leverage the strong supervised information obtained from cohesive pair-wises to pull samples into within-class, which avoid the similarity of high-dimension features located in different categories to achieve more the compact solution. We demonstrate the proposed method in popular datasets and compare the superiority with popular methods
半监督学习（Semi-supervised learning）是在模型构建视角下挖掘已标注和未标注样本之间潜在结构的一种强大范式。目前，基于图的模型解决了用空间度量直接表示样本分布的近似矩阵问题。其关键在于优化样本的连接，通过必须连接或不能连接来实现。遗憾的是，对于只有极少数标注样本的半监督聚类来说，要找到链接是相当困难的，因此在这种情况下会严重影响鲁棒性和准确性。为了解决这个问题，我们提出了内聚对智约束深度嵌入模型（Cohesive Pair-wises Constrained deep Embedding model，CPCE），以获得代表样本类别分布的最优嵌入，避免全局样本的图结构失效。CPCE 通过最小化样本重构误差来设计基于 CNN-Autoencoder 的深度网络框架，并同时对类内样本分布和类内类别分布建立约束，以优化潜在嵌入。然后，利用从内聚配对中获得的强监督信息，将样本拉入类内，从而避免了位于不同类别中的高维特征的相似性，实现了更紧凑的解决方案。我们在流行的数据集中演示了所提出的方法，并与流行的方法比较了其优越性
{"title":"Cohesive Pair-Wises Constrained Deep Embedding for Semi-Supervised Clustering with Very Few Labeled Samples*","authors":"Zhang Jing, Guiyan Wei, Yonggong Ren","doi":"10.34028/iajit/21/1/7","DOIUrl":"https://doi.org/10.34028/iajit/21/1/7","url":null,"abstract":"Semi-supervised learning is a powerful paradigm for excavating latent structures of between labeled and unlabeled samples under the view of models constructing. Currently, graph-based models solve the approximate matrix that directly represent distributions of samples by the spatial metric. The crux lies in optimizing connections of samples, which is achieved by subjecting to must-links or cannot-links. Unfortunately, to find links are rather difficult for semi-supervised clustering with very few labeled samples, therefore, significantly impairs the robustness and accuracy in such scenario. To address this problem, we propose the Cohesive Pair-wises Constrained deep Embedding model (CPCE) to obtain an optimal embedding for representing the category distribution of samples and avoid the failed graph-structure of the global samples. CPCE designs the deep network framework based on CNN-Autoencoder by minimizing reconstruct errors of samples, and build up constrains both of the sample distribution for within-class and the category distribution for intra-class to optimal the latent embedding. Then, leverage the strong supervised information obtained from cohesive pair-wises to pull samples into within-class, which avoid the similarity of high-dimension features located in different categories to achieve more the compact solution. We demonstrate the proposed method in popular datasets and compare the superiority with popular methods","PeriodicalId":161392,"journal":{"name":"The International Arab Journal of Information Technology","volume":"85 21","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139125135","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}
Computer vision enables to detect many objects in any scenario which helps in various real time application but still face recognition and detection remains a tedious process due to the low resolution, blurriness, noise, diverse pose and expression and occlusions. This proposal develops a novel scrupulous Standardized Convolute Generative Adversarial Network (SCGAN) framework for performing accurate face recognition automatically by restoring the occluded region including blind face restoration. Initially, a scrupulous image refining technique is utilised to offer the appropriate input to the network in the subsequent process. Following the pre-processing stage, a Caffe based Principle Component Analysis (PCA) filtration is conducted which uses convolutional architecture for fast feature embedding that collects spatial information and significant differentiating characteristics to counteract the loss of information existing in pooling operations. Then a filtration method identifies the specific match of the face based on the extracted features, creating uncorrelated variables that optimise variance across time while minimising information loss. To handle all the diversification occurring in the image and accurately recognise the face with occlusion in any part of the face, a novel Standardized Convolute GAN network is used to restore the image and recognise the face using novel Generative Adversarial Network (GAN) networks are modelled. This GAN ensures the normal distribution along with parametric optimization contributing to the high performance with accuracy of 96.05% and Peak Signal to Noise Ratio (PSNR) of 18 and Structural Similarity Index Metric (SSIM) of 98% for restored face recognition. Thus, the performance of the framework based on properly recognizing the face from the generated images is evaluated and discussed.
计算机视觉能够检测任何场景中的许多物体，这有助于各种实时应用，但由于分辨率低、模糊、噪声、姿势和表情各异以及遮挡等原因，人脸识别和检测仍然是一个繁琐的过程。本提案开发了一种新颖的无差别标准化卷积生成对抗网络（SCGAN）框架，通过恢复遮挡区域（包括盲目的人脸修复）自动执行准确的人脸识别。首先，利用严格的图像细化技术，在后续过程中为网络提供适当的输入。在预处理阶段之后，进行基于 Caffe 的主成分分析（PCA）过滤，该过滤使用卷积架构进行快速特征嵌入，收集空间信息和重要的区分特征，以抵消池化操作中存在的信息损失。然后，过滤方法根据提取的特征识别人脸的特定匹配，创建不相关的变量，优化跨时间的方差，同时最大限度地减少信息丢失。为了处理图像中出现的所有多样化情况，并在人脸的任何部分出现闭塞的情况下准确识别人脸，使用了一种新颖的标准化卷积 GAN 网络来还原图像，并使用新颖的生成对抗网络 (GAN) 网络模型来识别人脸。该 GAN 网络确保了正态分布，并对参数进行了优化，从而使还原的人脸识别准确率达到 96.05%，峰值信噪比（PSNR）达到 18%，结构相似度指标（SSIM）达到 98%。因此，我们对基于从生成的图像中正确识别人脸的框架的性能进行了评估和讨论。
{"title":"Scrupulous SCGAN Framework for Recognition of Restored Images with Caffe based PCA Filtration","authors":"Khushboo Agarwal, Manish Dixit","doi":"10.34028//iajit/21/1/10","DOIUrl":"https://doi.org/10.34028//iajit/21/1/10","url":null,"abstract":"Computer vision enables to detect many objects in any scenario which helps in various real time application but still face recognition and detection remains a tedious process due to the low resolution, blurriness, noise, diverse pose and expression and occlusions. This proposal develops a novel scrupulous Standardized Convolute Generative Adversarial Network (SCGAN) framework for performing accurate face recognition automatically by restoring the occluded region including blind face restoration. Initially, a scrupulous image refining technique is utilised to offer the appropriate input to the network in the subsequent process. Following the pre-processing stage, a Caffe based Principle Component Analysis (PCA) filtration is conducted which uses convolutional architecture for fast feature embedding that collects spatial information and significant differentiating characteristics to counteract the loss of information existing in pooling operations. Then a filtration method identifies the specific match of the face based on the extracted features, creating uncorrelated variables that optimise variance across time while minimising information loss. To handle all the diversification occurring in the image and accurately recognise the face with occlusion in any part of the face, a novel Standardized Convolute GAN network is used to restore the image and recognise the face using novel Generative Adversarial Network (GAN) networks are modelled. This GAN ensures the normal distribution along with parametric optimization contributing to the high performance with accuracy of 96.05% and Peak Signal to Noise Ratio (PSNR) of 18 and Structural Similarity Index Metric (SSIM) of 98% for restored face recognition. Thus, the performance of the framework based on properly recognizing the face from the generated images is evaluated and discussed.","PeriodicalId":161392,"journal":{"name":"The International Arab Journal of Information Technology","volume":"39 18","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139125635","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}
Now-a-days, healthcare monitoring system is very much important system in the medical field to know the patient‘s health status immediately. In the proposed system, the sensors are fixed over the patient’s body or placed at some distances around the body to collect patient’s significant parameters like blood pressure, temperature, heart beat rate, etc. These parameters are collected by the healthcare professionals through some connectivity mechanisms like Bluetooth, ZigBee, etc. These significant data will be outsourced to the cloud storage in a secure way to avoid attack from the attackers. So, we need some protection mechanism to safeguard this information. This article proposes a light weight cryptographic algorithm (symmetric key) via random number key generation using Diffie-Hellman key exchange based on Elliptic Curve (ECDH) cryptography. As a result of substituting bytes (S-box) and folding (horizontal and vertical) operations, the proposed symmetric key algorithm achieves the foremost properties of cryptography such as confusion and diffusion quite well. Experimental results showed that the overall execution time of the proposed algorithm is superior to the standard Advanced Encryption Standard (AES) algorithm. The throughput rate of the proposed algorithm is 20.525095 KB/seconds whereas for the standard AES algorithm throughput rate is 18.727215 KB/seconds. So, the proposed algorithm is faster than the existing AES algorithm. Moreover, the construction of S-box, IS-box and the key generation procedures are entirely different in the proposed algorithm so, it increases the complexity for the attacker and it will create confusion to the attacker
{"title":"Healthcare Data Security in Cloud Storage Using Light Weight Symmetric Key Algorithm","authors":"Vimala Devi Parthasarathy, Kalaichelvi Viswalingam","doi":"10.34028/iajit/21/1/5","DOIUrl":"https://doi.org/10.34028/iajit/21/1/5","url":null,"abstract":"Now-a-days, healthcare monitoring system is very much important system in the medical field to know the patient‘s health status immediately. In the proposed system, the sensors are fixed over the patient’s body or placed at some distances around the body to collect patient’s significant parameters like blood pressure, temperature, heart beat rate, etc. These parameters are collected by the healthcare professionals through some connectivity mechanisms like Bluetooth, ZigBee, etc. These significant data will be outsourced to the cloud storage in a secure way to avoid attack from the attackers. So, we need some protection mechanism to safeguard this information. This article proposes a light weight cryptographic algorithm (symmetric key) via random number key generation using Diffie-Hellman key exchange based on Elliptic Curve (ECDH) cryptography. As a result of substituting bytes (S-box) and folding (horizontal and vertical) operations, the proposed symmetric key algorithm achieves the foremost properties of cryptography such as confusion and diffusion quite well. Experimental results showed that the overall execution time of the proposed algorithm is superior to the standard Advanced Encryption Standard (AES) algorithm. The throughput rate of the proposed algorithm is 20.525095 KB/seconds whereas for the standard AES algorithm throughput rate is 18.727215 KB/seconds. So, the proposed algorithm is faster than the existing AES algorithm. Moreover, the construction of S-box, IS-box and the key generation procedures are entirely different in the proposed algorithm so, it increases the complexity for the attacker and it will create confusion to the attacker","PeriodicalId":161392,"journal":{"name":"The International Arab Journal of Information Technology","volume":"60 9","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139126903","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}
In Cloud Computing (CC) environment, requests of user are maintained via workloads that are allocated to Virtual Machines (VMs) using scheduling techniques which primarily focus on reducing the time for processing by generating efficient schedules of smaller lengths. The efficient processing of requests also needs larger usage of resources that incurs higher overhead in the form of utilization of energy and optimization of cost utilized by Physical Machines (PMs). Assignment of VMs optimally in the environment of CC for jobs submitted by users is a challenge. In order to obtain better solution involving scheduling of jobs to VMs, considering two parameters utilization of energy and cost, we present a dual-objective approach for VM allocation with improved scheduling of jobs in CC environment. The proposed work aimed to build a dual-objective scheduling model for improved job scheduling, focusing on minimization of cost and utilization of energy at a time. For evaluating performance of dual-objective approach, we utilized two types of benchmark datasets and compared with existing approaches such as Whale, Artificial Bee Colony (ABC), Particle Swarm Optimization (PSO) and Metaheuristic Dynamic VM Allocation (MDVMA) techniques. The results obtained from simulation demonstrated that dual-objective approach performs better in the form of minimization of utilization of energy and cost
在云计算（CC）环境中，用户的请求通过使用调度技术分配给虚拟机（VM）的工作负载来维持，调度技术主要侧重于通过生成长度较小的高效调度来减少处理时间。高效处理请求还需要使用更多的资源，从而产生更高的能源利用率和物理机（PM）成本优化形式的开销。在 CC 环境中为用户提交的任务优化分配虚拟机是一项挑战。为了在考虑能源利用率和成本这两个参数的情况下获得更好的解决方案，我们提出了一种双目标方法，用于在 CC 环境中改进作业调度的虚拟机分配。所提出的工作旨在为改进作业调度建立一个双目标调度模型，同时关注成本最小化和能源利用率。为了评估双目标方法的性能，我们使用了两种类型的基准数据集，并与现有的方法进行了比较，如 Whale、人工蜂群（ABC）、粒子群优化（PSO）和元启发式动态虚拟机分配（MDVMA）技术。模拟结果表明，双目标方法在能源利用率和成本最小化方面表现更好。
{"title":"A Dual-Objective Approach for Allocation of Virtual Machine with improved Job Scheduling in Cloud Computing","authors":"Sandeep Sutar, Manjunathswamy Byranahallieraiah, Kumarswamy Shivashankaraiah","doi":"10.34028/iajit/21/1/4","DOIUrl":"https://doi.org/10.34028/iajit/21/1/4","url":null,"abstract":"In Cloud Computing (CC) environment, requests of user are maintained via workloads that are allocated to Virtual Machines (VMs) using scheduling techniques which primarily focus on reducing the time for processing by generating efficient schedules of smaller lengths. The efficient processing of requests also needs larger usage of resources that incurs higher overhead in the form of utilization of energy and optimization of cost utilized by Physical Machines (PMs). Assignment of VMs optimally in the environment of CC for jobs submitted by users is a challenge. In order to obtain better solution involving scheduling of jobs to VMs, considering two parameters utilization of energy and cost, we present a dual-objective approach for VM allocation with improved scheduling of jobs in CC environment. The proposed work aimed to build a dual-objective scheduling model for improved job scheduling, focusing on minimization of cost and utilization of energy at a time. For evaluating performance of dual-objective approach, we utilized two types of benchmark datasets and compared with existing approaches such as Whale, Artificial Bee Colony (ABC), Particle Swarm Optimization (PSO) and Metaheuristic Dynamic VM Allocation (MDVMA) techniques. The results obtained from simulation demonstrated that dual-objective approach performs better in the form of minimization of utilization of energy and cost","PeriodicalId":161392,"journal":{"name":"The International Arab Journal of Information Technology","volume":"22 20","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139127187","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：