检索结果-内蒙古大学图书馆

Improving Generalization for Hyperspectral Image Classification:The Impact of Disjoint Sampling on Deep Models

computers, Materials & Continua 2024年第10期81卷 503-532页

作者： Muhammad Ahmad Manuel Mazzara Salvatore Distefano Adil Mehmood Khan Hamad Ahmed Altuwaijri Department of Computer Science National University of Computer and Emerging SciencesChiniot35400Pakistan Institute of Software Development and Engineering Innopolis UniversityInnopolis420500Russia Dipartimento di Matematica e Informatica—MIFT University of MessinaMessina98121Italy School of Computer Science University of HullHullHU67RXUK Department of Geography College of Humanities and Social SciencesKing Saud UniversityRiyadh11451Saudi Arabia

Disjoint sampling is critical for rigorous and unbiased evaluation of state-of-the-art(SOTA)models e.g.,Attention Graph and Vision *** training,validation,and test sets overlap or share data,it introduces a bias that inflates performance metrics and prevents accurate assessment of a model’s true ability to generalize to new *** paper presents an innovative disjoint sampling approach for training SOTA models for the Hyperspectral Image Classification(HSIC).By separating training,validation,and test data without overlap,the proposed method facilitates a fairer evaluation of how well a model can classify pixels it was not exposed to during training or *** demonstrate the approach significantly improves a model’s generalization compared to alternatives that include training and validation data in test data(A trivial approach involves testing the model on the entire Hyperspectral dataset to generate the ground truth *** approach produces higher accuracy but ultimately results in low generalization performance).Disjoint sampling eliminates data leakage between sets and provides reliable metrics for benchmarking progress in *** sampling is critical for advancing SOTA models and their real-world application to large-scale land mapping with Hyperspectral ***,with the disjoint test set,the performance of the deep models achieves 96.36%accuracy on Indian Pines data,99.73%on Pavia University data,98.29%on University of Houston data,99.43%on Botswana data,and 99.88%on Salinas data.

关键词： Hyperspectral image classification disjoint sampling Graph CNN spatial-spectral transformer

来源：评论

学校读者我要写书评

暂无评论

MASKED DISTILLATION WITH RECEPTIVE TOKENS 11

MASKED DISTILLATION WITH RECEPTIVE TOKENS

引用

11th International Conference on Learning Representations, ICLR 2023

作者： Huang, Tao Zhang, Yuan You, Shan Wang, Fei Qian, Chen Cao, Jian Xu, Chang School of Computer Science Faculty of Engineering The University of Sydney Australia SenseTime Research China School of Software and Microelectronics Peking University China University of Science and Technology of China China

Distilling from the feature maps can be fairly effective for dense prediction tasks since both the feature discriminability and localization priors can be well ***, not every pixel contributes equally to the performance, and a good student should learn from what really matters to the *** this paper, we introduce a learnable embedding dubbed receptive token to localize those pixels of interests (PoIs) in the feature map, with a distillation mask generated via pixel-wise *** the distillation will be performed on the mask via pixel-wise *** this way, a distillation mask actually indicates a pattern of pixel dependencies within feature maps of *** thus adopt multiple receptive tokens to investigate more sophisticated and informative pixel dependencies to further enhance the *** obtain a group of masks, the receptive tokens are learned via the regular task loss but with teacher fixed, and we also leverage a Dice loss to enrich the diversity of learned *** method dubbed MasKD is simple and practical, and needs no priors of tasks in *** show that our MasKD can achieve state-of-the-art performance consistently on object detection and semantic segmentation *** is available at https://***/hunto/MasKD. © 2023 11th International Conference on Learning Representations, ICLR 2023. All rights reserved.

关键词： Distillation

来源：评论

学校读者我要写书评

暂无评论

PUF-Based Privacy-Preserving Simultaneous Authentication Among Multiple Vehicles in VANET

引用

IEEE Transactions on Vehicular Technology 2024年第5期73卷 6727-6739页

作者： Awais, Syed Muhammad Yucheng, Wu Mahmood, Khalid Akram, Muhammad Wahid Hussain, Shafiq Das, Ashok Kumar Park, Youngho Chongqing University School of Microelectronics and Communication Engineering Chongqing400044 China National Yunlin University of Science and Technology Graduate School of Intelligent Data Science Yunlin64002 Taiwan University of Lahore Department of Software Engineering Punjab Lahore54590 Pakistan University of Sahiwal Department of Computer Science Punjab Sahiwal57000 Pakistan International Institute of Information Technology Center for Security Theory and Algorithmic Research Telangana Hyderabad500032 India Kyungpook National University School of Electronics Engineering Daegu41566 Korea Republic of

The intellectual and immense attainment of a smart city and auto-mobile vehicles are gradually growing. This growth drives Vehicular Ad-Hoc Network (VANET), a broadly operated complex network for inter-vehicular transmission to acquire information relevant to the vehicle location, road condition, traffic congestion, and speed. Generally, this information is procured through an open-source communication, which is susceptible to various security threats. In broad terms, the privacy of data in VANET is considered an imperative task. In this regard, many authentication protocols are formerly initiated for VANET. Most of these protocols are precarious owing to eavesdropping, replay, and impersonation attacks, etc. Furthermore, most existing protocols operates on enormous communication and computation rate, which turns them inadequate for VANET. This paper exhibits a lightweight privacy-preserving authentication protocol for n number of vehicles in VANET using the physical unclonable function (PUF) technique, and offers efficient and secure communication via public channels. Moreover, we evaluate our protocol formally and informally, showing that it is scalable, lightweight, and safe compared to formerly protocols. The rate of computation and communication is less than most related protocols. We also simulate the proposed scheme using the NS-3 simulator to measure its various network performance parameters. As a result, it is apparent through the beheld statistics that the presented protocol is beyond practical, efficient, and reliable for the secure dissemination of information in VANET. © 1967-2012 IEEE.

关键词： Vehicular ad hoc networks

来源：评论

学校读者我要写书评

暂无评论

Privacy-Enhanced Data Sharing in Multi-Institutional Healthcare

Privacy-Enhanced Data Sharing in Multi-Institutional Healthc...

引用

IEEE International Conference on Cloud Computing and Big Data Analysis (ICCCBDA)

作者： Jixiang Xiao Shengke Zeng Tiantian Zhou Yawen Feng School of Computer and Software Engineering Xihua University Chengdu China School of Computer Science and Engineering University of Electronic Science and Technology of China School of Computer and Software Engineering Xihua University Chengdu China

ISBN: (数字)9798350373554

ISBN: (纸本)9798350373561

In multi-institutional patient data sharing scenarios, maintaining fine-grained access control while safeguarding privacy and adapting to real-world environments is crucial. Traditional attribute-based encryption (ABE) schemes may inadvertently expose sensitive information through access policies and lack support for dynamic policy updates. To tackle these challenges, this paper proposes an ABE scheme with partially hidden access policies to bolster privacy protection for encrypted data stored on servers across healthcare institutions. Our scheme also facilitates access policy updates, ensuring accurate and efficient adaptation to evolving real-world needs. Through rigorous security analysis, we validate the efficacy of our approach in safeguarding patient data integrity, confidentiality, and adaptability to dynamic environments. Our scheme offers a robust solution for secure and efficient multi-institutional patient data sharing, supporting fine-grained access control, privacy preservation, and dynamic policy updates.

关键词： Access control Data privacy Cloud computing Data integrity Refining Medical services Encryption

来源：评论

学校读者我要写书评

暂无评论

A Study of Data Augmentation for Imbalance Problem in Network Traffic 3

A Study of Data Augmentation for Imbalance Problem in Networ...

引用

3rd International Conference on Blockchain Technology and Information Security, ICBCTIS 2023

作者： Wang, Xu Sun, Shimin Han, Li School of Computer Science and Engineering Tianjin University of Technology Tianjin China Tianjin Key Laboratory of Autonomous Intelligent Technology and System School of Software Tiangong University Tianjin China School of Computer Science and Engineering Tianjin University of Technology National Engineering Laboratory for Computer Virus Prevention and Control Technology Tianjin China

ISBN: (纸本)9798350340402

The complexity and diversity of smart grid systems increase the likelihood of anomalies in communications between devices in the system, and how these anomalies are detected is critical to the security of the grid system. However, the distribution of traffic types in the grid is often highly unbalanced, with the total amount of normal traffic data often far outweighing the anomalous data. This imbalance compromises the effectiveness of the detection model and poses a significant threat to the overall security of the grid This paper improves on traditional conditional generation adversarial neural networks and designs a new approach to overcome this obstacle and enhance the security of smart grid systems. In the generator part, the spatial features of the data are captured by convolutional neural networks, and the temporal features of the data are captured by gated recurrent units. By integrating these two components into the generative network we can generate synthetic data with both temporal and spatial features, thus enhancing the minority class samples. By incorporating the augmented minority data into the training process, the proposed approach enables the detection model to understand the distribution of features of the minority anomaly samples more accurately. This improved learning accuracy translates into improved model performance, resulting in higher detection rates for Previously under-represented anomalies. In addition, our approach effectively addresses the inherent bias towards traffic types with higher flows typically observed in asymmetric datasets, thus providing a more equitable and comprehensive security solution for smart grid systems. © 2023 IEEE.

关键词： Anomaly detection

来源：评论

学校读者我要写书评

暂无评论

CIFD: A Distance for Complex Intuitionistic Fuzzy Set 4th

CIFD: A Distance for Complex Intuitionistic Fuzzy Set

引用

4th International Conference on Machine Learning for Cyber Security, ML4CS 2022

作者： Zhao, Yangyang Xiao, Fuyuan School of Computer and Information Science Southwest University Chongqing400715 China School of Big Data and Software Engineering Chongqing University Chongqing401331 China

ISBN: (纸本)9783031200953

Intuitionistic fuzzy set (IFS) has attracted much attention because it can deal with fuzziness and uncertainty more flexibly than traditional fuzzy set. Complex intuitionistic fuzzy set (CIFS) extends intuitionistic fuzzy to the complex plane, which can better express and process uncertain information. In this paper, a novel distance measure complex intuitionistic fuzzy distance (CIFD) is proposed for CIFSs. Firstly, inspired by Tanimoto coefficient, a new similarity measure between CIFSs is proposed. Then, based on the similarity measure, the CIFD is proposed and its non-negativity, non-degeneracy, and symmetry are analyzed. Moreover, when CIFS degenerates into classical IFS, CIFD is also applicable to measure the differences between IFSs. Finally, to illustrate the effectiveness of CIFD, an example is given at the end. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.

关键词： Fuzzy sets

来源：评论

学校读者我要写书评

暂无评论

Crack Detection Method for Asphalt Pavement Combining Multi-scale Features and Attention Mechanism 4

Crack Detection Method for Asphalt Pavement Combining Multi-...

引用

4th International Conference on Advances in Electrical engineering and computer Applications, AEECA 2023

作者： Xu, Zijian Wang, Lili Chongqing University of Posts and Telecommunications School of Computer Science and Technology Chongqing China Chongqing University of Posts and Telecommunications School of Software Engineering Chongqing China

ISBN: (纸本)9798350308082

Crack, as one of the common diseases of asphalt pavements, seriously affects the health of asphalt pavements. To cope with the demand of crack detection in the context of complex pavements, an improved network model with an encoder-decoder structure is proposed. First, dense long and short connections are combined with full-scale jump connections, and the jump connection structure yields fullscale feature information to each node of the decoding layer. Second, spatial and channel attention modules are incorporated into the proposed network. The former is used at the low level of the network to improve the ability to capture crack detail information, and the latter is applied at the high level of the proposed network to obtain the semantic information. Finally, improve network performance by building deep supervision network. The proposed network is compared with on three datasets, DeepCrack, CFD, and Crack500, and the F-score reaches 86.79%. In this paper, the network is effective in crack detection and plays a certain role in maintaining road safety. © 2023 IEEE.

关键词： Decoding

来源：评论

学校读者我要写书评

暂无评论

Deep Autoencoding One-Class time Series Anomaly Detection 48

Deep Autoencoding One-Class time Series Anomaly Detection

引用

48th IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2023

作者： Mou, Xudong Wang, Rui Wang, Tiejun Sun, Jie Li, Bo Wo, Tianyu Liu, Xudong Beihang University School of Computer Science and Engineering Beijing China Zhongguancun Laboratory Beijing China Beihang University Sklsde School of Software Beijing China

ISBN: (纸本)9781728163277

Time-series Anomaly Detection(AD) is widely used in monitoring and security applications in various industries and has become a hot spot in the field of deep learning. Normality-representation-based methods perform well in certain scenarios but may ignore some aspects of the overall normality. Feature-extraction-based methods always take a process of pre-training, whose target differs from AD, leading to a decline in AD performance. In this paper, we propose a new AD method called deep Autoencoding One-Class (AOC), which learns features with AutoEncoder(AE). Meanwhile, the normal context vectors from AE are constrained into a hypersphere small enough, similar to one-class methods. With an objective function that optimizes the two assumptions simultaneously, AOC learns various aspects of normality, which is more effective for AD. Experiments on public datasets show that our method outperforms existing baseline approaches. © 2023 IEEE.

关键词： Autoen-coder One-class classification Time Series Anomaly Detection

来源：评论

学校读者我要写书评

暂无评论

Tencent pretrain: A scalable and flexible toolkit for pre-Training models of different modalities 61

Tencent pretrain: A scalable and flexible toolkit for pre-Tr...

引用

61st Annual Meeting of the Association for Computational Linguistics, ACL 2023

作者： Zhao, Zhe Li, Yudong Hou, Cheng Zhao, Jing Tian, Rong Liu, Weijie Chen, Yiren Sun, Ningyuan Liu, Haoyan Mao, Weiquan Guo, Han Guo, Weigang Wu, Taiqiang Zhu, Tao Shi, Wenhang Chen, Chen Huang, Shan Chen, Sihong Liu, Liqun Li, Feifei Chen, Xiaoshuai Sun, Xingwu Kang, Zhanhui Du, Xiaoyong Shen, Linlin Yan, Kimmo Tencent AI Lab School of Computer Science and Software Engineering Shenzhen University China School of Information and DEKE MOE Renmin University of China China

ISBN: (纸本)9781959429708

Recently, the success of pre-Training in text domain has been fully extended to vision, audio, and cross-modal scenarios. The proposed pre-Training models of different modalities are showing a rising trend of homogeneity in their model structures, which brings the opportunity to implement different pre-Training models within a uniform framework. In this paper, we present TencentPretrain, a toolkit supporting pre-Training models of different modalities. The core feature of TencentPretrain is the modular design. The toolkit uniformly divides pretraining models into 5 components: embedding, encoder, target embedding, decoder, and target. As almost all of common modules are provided in each component, users can choose the desired modules from different components to build a complete pre-Training model. The modular design enables users to efficiently reproduce existing pre-Training models or build brand-new one. We test the toolkit on text, vision, and audio benchmarks and show that it can match the performance of the original implementations. © ACL-DEMO 2023. All rights reserved.

关键词： Embeddings

来源：评论

学校读者我要写书评

暂无评论

Reseed Skipping of Linear Congruential Generator for Multi-level Image Steganography Security of Financial Data

引用

SN computer science 2023年第2期4卷 124页

作者： Danso, Juliana Mantebea Missah, Yaw Marfo Gyamfi, Enoch Opanin Dankwa, Stephen Kwabena, Sarpong Department of Computer Science Kwame Nkrumah University of Science and Technology (KNUST) Kumasi Ghana School of Computer Science and Engineering University of Electronic Science and Technology of China Chengdu China School of Information and Software Engineering University of Electronic Science and Technology of China Chengdu China Department of Cyber Security and Computer Engineering Technology School of Computing and Information Sciences C.K. Tedam University of Technology and Applied Sciences Upper East Region Navrongo Ghana School of Automation Engineering University of Electronic Science and Technology of China Chengdu China

This paper proposed a multi-level image steganography model that integrates with image and plaintext encryption schemes. The model starts by partitioning the cover-image pixel positions into two regions of respective purposes, one is reserved to hold the data, and through the implementation of the dynamic difference expansion principle, the other sloughs some parts to the reserved region in responsive to the size of the data to embed. Continuously reseeding and initializing the parameter values of a Linear Congruential Generator (LCG), generate a set of random pixel positions as members of the reserved region. Before data embedding, the region reserved to hold the data as well as the data bits to embed, are then encrypted using AES with respective keys. Encryption keys are randomly generated using a Cryptographically Secure Pseudo-Random Number Generator (CSPRNG), the Fortuna algorithm. Multiplex security nature of using a randomization generator to reserve image region for encryption and encrypting plaintext data before performing steganography still ensures stability in computational complexities better than a logarithmic O(n log n). A visual inspection by evaluators who made 3000 choices to correctly differentiate stego images from its cover-images shows that only 47 choices were correct, 1884 choices were wrong, while 1069 choices were uncertain. An experiment shows an average Embedding Rates (ERs) of 0.00109 bpp and an average PSNR of 72. This concludes, as our integrated steganography and image encryption which implements the dynamic difference expansion principle, outputs encrypted stego mages that can deceive the naked human eyes into identifying them as rather cover-images, its embedding rate is faster (i.e. ER < 0.1), suitably good, and ensures better imperceptibility of cover-image alterations. When comparing these experimental results with existing methods, our proposed model shows significant competitiveness. We even show real-life scenarios on how this

关键词： AES encryption scheme Dynamic difference expansion principle Linear congruential generator Pseudo-random number generator Reserving room before encryption Steganography

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：