检索结果-内蒙古大学图书馆

2025 IEEE International Conference on Consumer Electronics, ICCE 2025

作者： Chen, Xiu-Zhi Huang, Hsiao-Wei Chen, Yen-Lin Dept. of Computer Science & Information Engineering National Taipei University of Technology Taipei Taiwan

ISBN: (纸本)9798331521165

Two approaches are available for WSOD: MIL and CAM. MIL-based methods are capable of addressing multiple targets and multiple class problems. However, these methods require the input of region proposals. In contrast, CAM-based methods are straightforward and more efficient, yet they are constrained by limited localization accuracy and are likely to be applicable only to single target classes. This study proposes a CAM-based WSOD method that is capable of addressing multi-object, multi-class tasks with high efficiency, which could prove beneficial for the development of specific consumer electronic applications, such as smart cameras and smart home devices. The proposed method was evaluated on two public datasets, demonstrating that it is significantly more efficient than the baseline method. © 2025 IEEE.

关键词： Object detection

来源：评论

学校读者我要写书评

暂无评论

Deep Learning-Driven Cloud Intrusion Detection: Internet of Things and Cloud Networks 1

Deep Learning-Driven Cloud Intrusion Detection: Internet of ...

引用

1st International Conference on Advances in computer science, Electrical, Electronics, and Communication Technologies, CE2CT 2025

作者： Moon, Iffath Tanjim Islam Mondal, Md. Nazrul Rajshahi University of Engineering & Technology Dept. of Computer Science & Engineering Rajshahi6204 Bangladesh

ISBN: (纸本)9798331518578

In this paper, we have utilized deep learning approaches to detect cloud intrusion for the Internet of Things (IoT). The emerging growth of IoT and cloud environments has revolutionized many industries by allowing real-time data sharing and storage. Cloud platforms offer the essential infrastructure for managing the massive volumes of data produced by IoT devices. However, due to several flaws, including insufficient encryption, insecure APIs, and weak authentication, intruders can exploit the vulnerabilities to interrupt services and obtain unauthorized access. Therefore, deep learning can be an effective solution to detect cloud intrusion in IoT. In this research, we have developed individual convolutional neural network (CNN) and long short-term memory (LSTM) models, along with a hybrid CNN-LSTM model for multiclass classification and binary classification to detect cloud intrusion. In addition, we have examined which models perform the best for cloud intrusion detection for IoT among the individual and hybrid models. The proposed models are effective in detecting cloud intrusion, as CNN can extract spatial features, and LSTM can capture temporal dependencies from network traffic data. Furthermore, we have utilized one of the most recent and comprehensive datasets for IoT, the ToN-IoT dataset, containing real-world data. In multiclass classification, the hybrid CNN-LSTM has given the best performance, achieving an accuracy of 97.65%, precision of 97.72%, recall of 97.65%, f1 score of 97.67%, and AUC of 99.93%. In binary classification, all CNN, LSTM, and hybrid CNN-LSTM have demonstrated high performance, where LSTM has achieved the highest accuracy of 99.97%. Moreover, we have evaluated the performance of multiclass classification for detecting nine distinct kinds of attacks. In this instance, CNN has shown better performance in detecting backdoor attacks. Thus, this research can make the IoT platform more secure by efficiently detecting cloud intrusion. © 2025

关键词： Network intrusion

来源：评论

学校读者我要写书评

暂无评论

Quantum Cryptography and Post-Quantum Security: Safeguarding Cryptographic Protocols Against Quantum Threats 2

Quantum Cryptography and Post-Quantum Security: Safeguarding...

引用

2nd International Conference on Advanced Innovations in Smart Cities, ICAISC 2025

作者： Ishaq, Muhammad Al-Anesi, Bushra Jeddah International College Computer Science & Information Technology Dept. Jeddah Saudi Arabia

ISBN: (纸本)9798331506995

As quantum computing progresses towards practical realization, the security of existing cryptographic protocols, such as RSA and Elliptic Curve Cryptography (ECC), faces unprecedented risks. Quantum algorithms like Shor's Algorithm threaten to undermine the security of these widely used encryption schemes, necessitating the development of quantum-resistant alternatives. This paper explores the vulnerabilities of current cryptographic protocols in the quantum era and investigates quantum-safe cryptographic techniques such as lattice-based, hash-based, and code-based cryptography. Additionally, we delve into Quantum Key Distribution (QKD), a method that leverages the principles of quantum mechanics to securely distribute cryptographic keys, ensuring secure communication even in the presence of quantum adversaries. The integration of QKD with post-quantum cryptographic algorithms is proposed as a robust approach to future-proofing cybersecurity. The paper concludes with a discussion of practical challenges in implementing quantum-safe systems and highlights future research directions to strengthen cryptographic resilience against quantum threats. © 2025 IEEE.

关键词： Qubits

来源：评论

学校读者我要写书评

暂无评论

Reconsider Time Series Analysis for Insider Threat Detection

Reconsider Time Series Analysis for Insider Threat Detection

引用

2024 IEEE International Conference on Big Data, BigData 2024

作者： Chen, Chia-Cheng Pao, Hsing-Kuo National Taiwan University of Science and Technology Dept. of Computer Science and Information Engineering Taipei Taiwan

ISBN: (纸本)9798350362480

Insider threat detection (ITD) presents a significant challenge in cybersecurity, particularly within large and complex organizations. Traditionally, ITD has been overshadowed by the focus of external threats, resulting in less attention and development in this critical area. Conventional ITD approaches often rely heavily on event-driven approaches. On top of that, researchers developed various rule-based methods to conquer the tasks. Based on that, we often ignore the intrinsic temporal relationships that are naturally built in between events that occur in different moments. For instance, we may easily understand events with causality such as one anomalous event followed by another specific event to complete a malicious action;however, may not be aware of events that occur around 9 am every morning during working hours. In our opinion, we attempt to re-consider the temporal behavior to extract the information hidden in cyberspace activities. Specifically, some effective sentence embeddings can assist us in providing informative internal representations to summarize temporal behaviors in the temporal activity sequences to make the right judgment on insider threat detection. In this paper, we propose a novel methodology for insider threat detection that emphasizes temporal relationship modeling on top of already-matured event sequence analysis to effectively catch insider threats. The proposed approach leverages contrastive sentence embeddings to learn users' intentions in sequences, followed by the deployment of a user-level and event-level Contrastive Learning (euCL) model to incorporate temporal behaviors with user behavior embeddings. To validate the proposed methodology, we conduct extensive analyses and experiments using the publicly available CERT dataset. The results demonstrate the effectiveness and robustness of the proposed method in detecting insider threats and identifying malicious scenarios, highlighting its potential for enhancing cybersecurity measur

关键词： Contrastive Learning

来源：评论

学校读者我要写书评

暂无评论

Monocular UAV Motion Planning Method Based on Deep Reinforcement Learning

Monocular UAV Motion Planning Method Based on Deep Reinforce...

引用

2025 IEEE International Conference on Consumer Electronics, ICCE 2025

作者： Hsueh, Man-Chen Ku, Chun Chen, Xiu-Zhi Chen, Yen-Lin Dept. of Computer Science & Information Engineering National Taipei University of Technology Taipei Taiwan

ISBN: (纸本)9798331521165

This paper presents a monocular UAV motion planning method based on the DRL method SDDPG. By processing image data with the self-proposed lightweight dept. estimation model and a pretrained encoder, the information is sent into an improved Actor-Critic Network, which addresses overestimation issues and enhances learning efficiency. The experiments were conducted in Unreal Engine 4 and Airsim, which demonstrate the improvement of the motion planning ability, enabling UAVs to autonomously navigate and safely reach the defined destination. © 2025 IEEE.

关键词： Drones

来源：评论

学校读者我要写书评

暂无评论

Automated Skin Cancer Classification and Detection Using Convolutional Neural Networks and Dermoscopy Images 4

Automated Skin Cancer Classification and Detection Using Con...

引用

4th International Conference on Evolutionary Computing and Mobile Sustainable Networks, ICECMSN 2024

作者： Sara, Umme Das, Utpol Kanti Sikder, Juel Chowdhury, Sudipta Roy Dept. of Computer Science and Engineering National Institute of Textile Engineering and Research Savar Dhaka Bangladesh Dept. of Computer Science and Engineering Rangamati Science and Technology University Rangamati Sadar Rangamati4500 Bangladesh

The uncontrolled growth of skin cells in the epidermis producing the creation of a mass termed a tumor is a dangerous condition known as skin cancer. Current developments in deep learning artificial intelligence have greatly improved image-based diagnosis. In this study, we included a Skin Lesion Cancer feature extractor Convolutional Neural Network (SLC-CNN) model, which is used for both classification with the SVM classifier and segmentation with XGBoost for skin cancer. In our proposed system, a test image of skin cancer is taken and pre-processed for both classification and segmentation purposes. After applying pre-processing, the test image features are extracted using the SLC-CNN feature extractor, which features are used in SVM to classify the types of skin cancer (Benign and Malignant), and based on the classification result, a trained XGBoost model is called to segment the cancer region. We have tested our system using the dermoscopy image collection from the International Skin Imaging Collaboration (ISIC) and built it in Google Colab to best use the GPU. Our suggested approach has gained a segmentation accuracy of 95.25% and a classification accuracy of 99.6%. © 2025 The Authors. Published by Elsevier B.V.

关键词： Lung cancer

来源：评论

学校读者我要写书评

暂无评论

Fractional Delay and Doppler Estimation for OTFS systems with Doppler Squint Effect

Fractional Delay and Doppler Estimation for OTFS Systems wit...

引用

2025 IEEE Wireless Communications and Networking Conference, WCNC 2025

作者： Zhang, Meiying Cao, Ruoxiao Qian, Jun He, Hengtao Song, Shenghui Zhang, Jun Letaief, Khaled B. The Hong Kong University of Science and Technology Dept. of Electronic and Computer Engineering Hong Kong

ISBN: (纸本)9798350368369

Orthogonal time frequency space (OTFS) modulation is a promising technology for mitigating severe Doppler effects in high-mobility scenarios. However, existing OTFS channel estimation methods neglect the Doppler Squint Effect (DSE), which incurs serious performance loss. In this paper, we propose a channel estimation algorithm based on Newton's method to accurately estimate fractional delay and Doppler in OTFS systems with DSE. In particular, we obtain the maximum delay and Doppler grid spacing for codebook design to guarantee the convergence of the algorithm. Additionally, we derive the Cramér-Rao lower bound (CRLB) for testing channel parameter estimation performance of our proposed algorithm. Simulation results demonstrate that our proposed algorithm outperforms the orthogonal matching pursuit (OMP) algorithm in terms of normalized mean square error (NMSE), surpassing the Newtonized OMP algorithm with traditional dictionary matrix and approaching the CRLB performance. © 2025 IEEE.

关键词： Channel estimation

来源：评论

学校读者我要写书评

暂无评论

Extract, Define, Canonicalize: An LLM-based Framework for Knowledge Graph Construction

Extract, Define, Canonicalize: An LLM-based Framework for Kn...

引用

2024 Conference on Empirical Methods in Natural Language Processing, EMNLP 2024

作者： Zhang, Bowen Soh, Harold Dept. of Computer Science National University of Singapore Singapore NUS Smart Systems Institute Singapore

ISBN: (纸本)9798891761643

In this work, we are interested in automated methods for knowledge graph creation (KGC) from input text. Progress on large language models (LLMs) has prompted a series of recent works applying them to KGC, e.g., via zero/few-shot prompting. Despite successes on small domain-specific datasets, these models face difficulties scaling up to text common in many real-world applications. A principal issue is that, in prior methods, the KG schema has to be included in the LLM prompt to generate valid triplets;larger and more complex schemas easily exceed the LLMs' context window length. Furthermore, there are scenarios where a fixed pre-defined schema is not available and we would like the method to construct a high-quality KG with a succinct self-generated schema. To address these problems, we propose a three-phase framework named Extract-Define-Canonicalize (EDC): open information extraction followed by schema definition and post-hoc canonicalization. EDC is flexible in that it can be applied to settings where a pre-defined target schema is available and when it is not;in the latter case, it constructs a schema automatically and applies self-canonicalization. To further improve performance, we introduce a trained component that retrieves schema elements relevant to the input text;this improves the LLMs' extraction performance in a retrieval-augmented generation-like manner. We demonstrate on three KGC benchmarks that EDC is able to extract high-quality triplets without any parameter tuning and with significantly larger schemas compared to prior works. Code for EDC is available at https://***/clear-nus/edc. © 2024 Association for Computational Linguistics.

关键词： Knowledge graph

来源：评论

学校读者我要写书评

暂无评论

Advancements in Animal Tracking: Assessing Deep Learning Algorithms 15

Advancements in Animal Tracking: Assessing Deep Learning Alg...

引用

15th International Conference on Computing Communication and Networking Technologies, ICCCNT 2024

作者： Akter, Tahmina Majumder, Avishek Mahmud, Tanjim Habib, Israt Binteh Hossain, Mohammad Shahadat Andersson, Karl Port City International University Dept. of Computer Science and Engineering Chittagong Bangladesh Rangamati Science and Technology University Dept. of Computer Science and Engineering Rangamati4500 Bangladesh International Islamic University Chittagong Dept. of Computer Science and Engineering Chittagong Bangladesh University of Chittagong Dept. of Computer Science and Engineering Chittagong4331 Bangladesh Luleå University of Technology Cybersecurity Laboratory Luleå97187 Sweden

ISBN: (纸本)9798350370249

In the rapidly advancing field of computer vision, object detection has become crucial for various applications, including animal tracking, face detection, and surveillance systems. This study investigates the efficacy of contemporary object detection methodologies by evaluating the performance of the You Only Look Once (YOLO) models and TensorFlow Model Zoo architectures for animal tracking. YOLO models, known for their ability to process entire images in real-time and predict bounding boxes and class probabilities simultaneously, offer significant advantages over traditional methods such as Convolutional Neural Networks (CNNs) and Fast R-CNNs. This paper compares the performance of YOLOv5 and YOLOv7, alongside TensorFlow-based models like Faster R-CNN ResNetv152 and SSD ResNet101, using a dataset of animal images. Our findings reveal that YOLOv5 outperforms other models with a mean average precision (mAP) of 9 7.5%, demonstrating superior accuracy and efficiency in object detection tasks. YOLOv7 also shows strong performance with an mAP of 96.7%, while TensorFlow Model Zoo's Faster R-CNN and SSD models lag behind with mAPs of 81.9% and 81.6%, respectively. The results highlight the significant advancements in deep learning and object detection algorithms, particularly the advantages of YOLO's architecture in handling complex detection tasks in real-world scenarios. © 2024 IEEE.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Securing of Cloud Storage Data Using Hybrid AES-ECC Cryptographic Approach

引用

Journal of Mobile Multimedia 2023年第2期19卷 363-388页

作者： Kumar, Sunil Kumar, Dilip Dept. of Computer Science and Engineering National Institute of Technology Jamshedpur India

Internet has revolutionized the world in a way no one could have ever imagined. It paved the way for various different technologies, that have transformed the world exceptionally. Internet enabled cloud technology which provides cost-effective, scalable, on-demand computing resources with little to no downtime. Cloud storage allows its users to store and access private data from anywhere in the world without needing any high-end computing system. Cloud storage isn’t always secure, but that doesn’t imply it isn’t. The security of a data depends on the security policies followed by the provider along with the security of the communication channel via which the data is being sent. Encryption is used to obfuscate the data so that it can only be viewed when correct credentials, known as encryption keys, are provided. Following study proposes an encryption technique using ECC (Elliptic Curve Cryptography) along with AES (Advance Encryption Standard) to provide data confidentiality in an efficient way for securing data on cloud and hence, protect the personal information of user from any adversary. This new method is more effective, and the results are superior as a consequence. © 2022 River Publishers.

关键词： AES asymmetric encryption authentication blowfish cloud storage security ECC Symmetric encryption

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：