检索结果-内蒙古大学图书馆

2nd International Conference on Cyber Resilience, ICCR 2024

作者： Al-E'mari, Salam Sanjalawe, Yousef Alqudah, Hamzah University Of Petra Faculty Of Information Technology Information Security Department Amman Jordan American University Of Madaba Faculty Of Information Technology Cybersecurity Department Amman Jordan American University Of Madaba Faculty Of Information Technology Department Of Data Science And Artificial Intelligence Amman Jordan

ISBN: (纸本)9798350394962

This paper presents a novel approach to enhancing real-time surveillance systems using an augmented YOLO (You Only Look Once) v8 architecture. The primary focus is on integrating advanced security features into the YOLO framework to improve the detection and analysis of moving objects in various surveillance scenarios. The key enhancements include the implementation of an Anomaly Detection Layer, the integration of Behavior Analysis Algorithms, and the enhancement of Real-Time data Processing capabilities. These modifications are designed to increase the accuracy of threat detection while maintaining the high processing speeds essential for real-time applications. The Anomaly Detection Layer utilizes unsupervised learning algorithms to identify deviations from normal patterns, enabling the system to flag potential security threats that might otherwise go unnoticed. The Behavior Analysis Algorithms are incorporated to analyze movement patterns and identify suspicious behaviors, adding a layer of contextual understanding to object detection. To ensure the practicality of these enhancements, the model is optimized to handle live video feeds with minimal latency, maintaining an average processing speed of 45 frames per second. The experimental setup involved testing the enhanced YOLO v8 model in various simulated and real-world environments, comparing its performance with standard YOLO v8 and other existing surveillance models. The results demonstrated a significant improvement in detection accuracy, particularly in identifying anomalous behaviors and potential threats. The enhanced model outperformed the baseline models in various complex scenarios, including crowded public spaces and high-activity zones. This research contributes to the field of surveillance technology by offering a more sophisticated and efficient tool for public safety and security applications. It highlights the potential of integrating advanced AI features into existing object detection frameworks

关键词： Object detection

来源：评论

学校读者我要写书评

暂无评论

Latency- and Privacy-Aware Convolutional Neural Network Distributed Inference for Reliable Artificial Intelligence Systems

IEEE Transactions on Artificial Intelligence

引用

IEEE Transactions on Artificial Intelligence 2025年第2期6卷 365-377页

作者： Hu, Yuhao Xu, Xiaolong Qi, Lianyong Zhou, Xiaokang Xia, Xiaoyu Nanjing University of Information Science and Technology School of Software Nanjing210044 China College of computer science and technology Qingdao266850 China Shiga University Faculty of Data Science Hikone522-8522 Japan RMIT University School of Computing Technologies MelbourneVIC3000 Australia

Reliable artificial intelligence (AI) systems not only propose a challenge on providing intelligent services with high quality for customers but also require customers' privacy to be protected as much as possible during process of the services. Given the ultrahigh computing load brought by deep-learning-based intelligent services to edge devices and the ultralong distance between edge and cloud, low-latency requirement of intelligent services is hard to meet in single edge computing or cloud computing. Edge-cloud collaborative inference of deep neural networks (DNNs) is considered a feasible solution to the problem. However, former work has not reduced the inference latency to the greatest extent and has not considered privacy protection in distributed systems. To solve the problem, we first establish a novel queue mechanism. Then, the convolution layer split decisions are made based on deep reinforcement learning (DRL) to realize the parallel inference of convolutional neural networks (CNNs) for inference latency reduction. Next, for each CNN, the partition decision is made based on brute force algorithm to further reduce inference latency and protect customers' privacy. Finally, simulation results show that our method performs better than the existing other methods. © 2020 IEEE.

关键词： Cloud computing

来源：评论

学校读者我要写书评

暂无评论

Enhancing intrusion detection in IIoT: optimized CNN model with multi-class SMOTE balancing

引用

Neural Computing and Applications 2024年第24期36卷 14643-14659页

作者： Eid, Abdulrahman Mahmoud Soudan, Bassel Nassif, Ali Bou Injadat, MohammadNoor Department of Computer Engineering College of Computing and Informatics University of Sharjah Sharjah United Arab Emirates Department of Data Science and AI Faculty of Information Technology Zarqa University Zarqa Jordan

This work introduces an intrusion detection system (IDS) tailored for industrial internet of things (IIoT) environments based on an optimized convolutional neural network (CNN) model. The model is trained on a dataset that was balanced using a novel multi-class implementation of synthetic minority over-sampling technique (SMOTE) that ensures equal representation of all classes. Additionally, systematic optimization will be used to fine tune the hyperparameters of the CNN model and mitigate the effects of the increased size of the training dataset. Evaluation results will demonstrate substantial improvement in performance when the optimized CNN model is trained on the balanced dataset. The proposed IDS will be evaluated using the IIoT-specific WUSTL-IIOT-2021 dataset, and then its generalization capability will be verified using the non-domain specific UNSW_NB15 dataset. The model’s performance will be evaluated using accuracy, precision, recall, and F1-score metrics. The results will demonstrate that the proposed IDS is highly effective with performance exceeding 99.9% on all performance metrics. The IDS is also highly effective in detecting intrusion for generic IT networks achieving improvements in excess of 30% compared to the default baseline model. The results emphasize the versatility and effectiveness of the proposed IDS model, making it a reliable and adaptable solution for enhancing network security across diverse network environments. © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2024.

关键词： Neural network models

来源：评论

学校读者我要写书评

暂无评论

Feature Selection in Associative Classification-A Review and Comparative Analysis 24

Feature Selection in Associative Classification-A Review and...

引用

24th International Arab Conference on information technology, ACIT 2023

作者： Alazaidah, Raed Al-Qerem, Ahmad Qasem, Mais Haj Al-Shaikh, Ala'a Almilli, Nabeel Injadat, MohammadNoor Zarqa University Faculty of Information Technology Department of Data Science and Ai Zarqa Jordan Zarqa University Faculty of Information Technology Department of Computer Science Zarqa Jordan

ISBN: (纸本)9798350384307

Feature selection is one of the most important and significant steps that highly affects the final accuracy of any classification model. It aims to determine the most significant features that help in accurately predicting the class label for any new case. This research aims to survey and identify the best feature selection method that suits algorithms belong to Associative Classification (AC). Hence, nine different feature selection methods have been evaluated and compared using five AC based algorithms with respect to Accuracy metric. Results revealed that ABB-IEP and ABB-LIU methods show the best performance with datasets that contain nominal features only, while Relief feature selection method showed the best Accuracy with datasets that comprise Real and Integer features. © 2023 IEEE.

关键词： associative classification classification data mining data science feature selection pre-processing

来源：评论

学校读者我要写书评

暂无评论

A Comparative Analysis of Discretization Techniques in Machine Learning 24

A Comparative Analysis of Discretization Techniques in Machi...

引用

24th International Arab Conference on information technology, ACIT 2023

作者： Alazaidah, Raed Zarqa University Faculty of Information Technology Department of Data Science and Ai Zarqa Jordan

ISBN: (纸本)9798350384307

data discretization is a very significant pre-process step that has a great impact on the predictive performance of several machine learning algorithms. This step aims to convert continuous attribute with large number of real values into a finite set of intervals with minimum data loss. Several techniques have been proposed to accomplish this task, without a proof which one is better with respect to data characteristics. Hence, this paper investigates and evaluates fifteen discretization techniques using two popular Associative Classification (AC) based classifiers (CBA and CBA2), with respect to Accuracy metric. The results revealed that both Ameva and Chi-2 techniques show a superior performance using five different datasets comparing with the other considered techniques. © 2023 IEEE.

关键词： associative classification classification data mining data science discretization machine learning

来源：评论

学校读者我要写书评

暂无评论

Multi Label Classification Algorithm Based on Conditional Local Correlations among Labels and Ensemble Learning 25

Multi Label Classification Algorithm Based on Conditional Lo...

引用

25th International Arab Conference on information technology, ACIT 2024

作者： Alazaidah, Raed Nasir, Ola Hadi, Wael Samara, Ghassan Al Daoud, Essam Aljaidi, Mohammad Khouj, Mohammed Talat Elsoud, Esraa Abu Alzoubi, Haneen Zarqa University Faculty of Information Technology Department of Data Science and Ai Zarqa Jordan Zarqa University Faculty of Information Technology Department of Computer Science Zarqa Jordan Petra University Faculty of Information Technology Department of Cyber Security Amman Jordan University of Business and Technology Department of Electrical Engineering Jeddah21432 Saudi Arabia Applied Science Private University Faculty of Information Technology Department of Computer Science Amman Jordan

ISBN: (纸本)9798331540012

Multi-Label Classification (MLC) is an advanced form of traditional classification, where each data instance is simultaneously associated with multiple labels. The importance of MLC has increased significantly in recent years, due to its wide range of application areas. Modern algorithms, when faced with the challenge of multiple labels, continue to show suboptimal performance, conversely, they show significantly superior performance when processing single-label datasets. This paper proposes a new algorithm to deal with the problem of transforming datasets from multiple labels to single labels by exploiting conditional and local correlations between labels after applying the Apriori predictive algorithm to extract relationships. Six ensemble learning models: Voting Hard, Voting Soft, Voting Weighted, Stacking LR, Stacking RF, and Stacking SVM, along with four classification models: SMO, DecisionStump, Random Forest, and Random Committee, were applied to three datasets. The algorithm showed effective and good performance by comparing the results of the proposed Multi Label Classification algorithm based on Conditional Local Correlations among Labels (MLC-CLCL) with several popular algorithms for dealing with multiple labels. © 2024 IEEE.

关键词： Contrastive Learning

来源：评论

学校读者我要写书评

暂无评论

Enhanced accuracy of deep learning method for fruit images classification 25

Enhanced accuracy of deep learning method for fruit images c...

引用

25th International Arab Conference on information technology, ACIT 2024

作者： Arabiat, Mohammad Abuowaida, Suhaila Alshdaifat, Nawaf Aburomman, Ahmad Al Henawi, Esraa Dmour, Mohmmad Elrashidi, Ali Zarqa University Faculty of Information Technology Department of Computer Science Zarqa Jordan Al al-Bayt University Faculty of Information Technology Department of Computer Science Mafraq Jordan Applied Science Private University Faculty of Information Technology Amman Jordan Zarqa University Faculty of Information Technology Department of Data Science and Ai Zarqa Jordan University of Business and Technology Department of Electrical Engineering Jeddah21432 Saudi Arabia

ISBN: (纸本)9798331540012

In the last few years, artificial intelligence has shown its importance and ability in computer vision due to its high level of accuracy and wide acceptance. The main goal of the research is to look into a better way to separate different types of fruits. Although deep learning is used, the standard classification methods for different types of fruit don't work very well. The main steps of the recommended better multiple fruit classification method are: 1) Making changes to the filter's size. Selecting the most suitable number of repeats is part of the process of improving the ResNet-101 block. 3) It is proposed that there be one more convolutional layer. Exists support for the suggested method after collecting and labeling 27,000 pictures as part of an actual study. The dataset with many fruits is split into two parts: 80% is used for training and 20% is used for testing. These pictures were used to train an application based on deep learning (DL) to help successfully determine multiple fruit objects in images. Then and then possible to compare the results of this method to the most current methods in the area of object division. The application being studied at does better than expected in terms of accuracy AP in a range of levels. The values always go farther than the equivalent measures of three widely used methods. Compared to the other methods, the suggested one takes less time to train. There are higher numbers for the thresholds than there are for the state of the art method's thresholds © 2024 IEEE.

关键词： Deep reinforcement learning

来源：评论

学校读者我要写书评

暂无评论

Classifying Psychiatric Patients Using Machine Learning 25

Classifying Psychiatric Patients Using Machine Learning

引用

25th International Arab Conference on information technology, ACIT 2024

作者： Alrjoob, Hmmam M. Alazaidah, Raed Batyha, Radwan Khafajeh, Hayel Elsoud, Esraa Abu Saeb Al-Sherideh, Ala'a Elrashidi, Ali Alqaraleh, Muhyeeddin Amer, Waleed Zarqa University Faculty of Information Technology Department of Computer Science Zarqa Jordan Zarqa University Faculty of Information Technology Department of Data Science and Ai Zarqa Jordan Applied Science Private University Faculty of Information Technology Department of Computer Science Amman Jordan Zarqa University Faculty of Information Technology Department Computer Science Zarqa Jordan Zarqa University Faculty of Information Technology Department of Cyper Security Zarqa Jordan University of Business and Technology Department of Electrical Engineering Jeddah21432 Saudi Arabia Zarqa University Faculty of Information Technology Department of Software Engineering Zarqa Jordan Zarqa University Faculty on Information Technology Department of Cyber Security Zarqa Jordan

ISBN: (纸本)9798331540012

Psychological disorders are considered chronic illnesses that affect a wide range of populations. Some studies in the United States indicate that one in every eight individuals is affected by a psychological disorder. Additionally, individuals with a psychological disorder often have an impact on the nature of the lives of those around them. It is beneficial to detect the illness in its early stages to increase the likelihood of recovery. In this research, thirteen classification algorithms are used. Each algorithm has an interesting approach. It can be used to analyze several datasets where three datasets specific to Psychiatric patients were selected. The datasets have been chosen from UCI Machine Learning Repository. The datasets were re-analyzed after only 50% of the characteristics were approved after being evaluated by the several feature selection techniques. The main objective of this research is to find the best classifier and feature selection strategy in analyzing datasets of Psychiatric patients. After conducting numerous comparisons, it became evident that, the best classifier for the Autism-Adult-data dataset are Bagging and Random Forest. As for the best learning strategy, it was found to be Functions and Trees, the best feature selection was obtained using Symmetrical Uncourt Attribute Evaluation. Both IBK and KStar classifier proved to be the best classifier for the Parkinson's dataset, Lazy demonstrated to be the most effective learning strategy, Correlation Attribute Evaluation technique yielded the best feature selection. Logistic emerged as the best classifier for planning-relax dataset, the best learning strategy was Bayes, the best feature selection was found to be Symmetrical Uncert Feature Selection. © 2024 IEEE.

关键词： Contrastive Learning

来源：评论

学校读者我要写书评

暂无评论

GPMBC-FZ: A Query Optimization Method in Distributed database 6

GPMBC-FZ: A Query Optimization Method in Distributed Databas...

引用

6th International Conference on Next Generation data-Driven Networks, NGDN 2024

作者： Du, Yan Ding, Zhiming Faculty of Information Technology Beijing University of Technology Beijing China Center of Spatial-Temporal Data Management and Data Sciecne Institute of Software Chinese Academy of Science Beijing China

ISBN: (纸本)9798350388374

The rapid growth in the scale of distributed databases has led to a deeper impact of the multi-join query optimization problem on data query efficiency. The increase in the number of relations causes a significant increase in the search space complexity of distributed database multi-join optimization, which greatly affects the efficiency of finding the optimal execution plan, thus hindering the further optimization of multi-join queries. In this regard, a genetic multiple bee colony algorithm combined with fuzzy logic is proposed to solve the problem of further optimizing multi-join queries of distributed databases. Firstly, the fast convergence property of the genetic algorithm is utilized to obtain a set of relatively optimal solutions in order to overcome the blindness of the artificial bee algorithm's pre-search, and then the parallel multiple bee algorithm is introduced, which makes use of the information interaction between multiple bee colonies to better prevent the algorithm from falling into the local optimal situation as well as the parallelization of the execution of the multiple bee colonies to improve the search efficiency and convergence speed of the algorithm. Secondly, the fuzzy logic system is also used to optimize the parameters of the algorithm and set the corresponding algorithmic control parameters for different dimensional queries. Finally, a comparison experiment is conducted to test the average execution cost and algorithm execution time of the Top-K query plans output by the algorithm in the case of six query statements involving different lengths, and the experimental results show that the K optimal query plans generated by the method proposed in this paper have lower query excution cost and algorithm execution time, and can effectively improve the query efficiency. © 2024 IEEE.

关键词： Genetic algorithms

来源：评论

学校读者我要写书评

暂无评论

REINFORCING PRE-TRAINED MODELS USING COUNTERFACTUAL IMAGES 31

REINFORCING PRE-TRAINED MODELS USING COUNTERFACTUAL IMAGES

引用

31st IEEE International Conference on Image Processing, ICIP 2024

作者： Li, Xiang Togo, Ren Maeda, Keisuke Ogawa, Takahiro Haseyama, Miki Graduate School of Information Science and Technology Hokkaido University Japan Faculty of Information Science and Technology Hokkaido University Japan Data-Driven Interdisciplinary Research Emergence Department Hokkaido University Japan

ISBN: (纸本)9798350349399

This paper proposes a novel framework to reinforce classification models using language-guided generated counterfactual images. Deep learning classification models are often trained using datasets that mirror real-world scenarios. In this training process, because learning is based solely on correlations with labels, there is a risk that models may learn spurious relationships, such as an overreliance on features not central to the subject, like background elements in images. However, due to the black-box nature of the decision-making process in deep learning models, identifying and addressing these vulnerabilities has been particularly challenging. We introduce a novel framework for reinforcing the classification models, which consists of a two-stage process. First, we identify model weaknesses by testing the model using the counterfactual image dataset, which is generated by perturbed image captions. Subsequently, we employ the counterfactual images as an augmented dataset to fine-tune and reinforce the classification model. Through extensive experiments on several classification models across various datasets, we revealed that fine-tuning with a small set of counterfactual images effectively strengthens the model. © 2024 IEEE.

关键词： Deep reinforcement learning

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：