检索结果-内蒙古大学图书馆

International Conference on Acoustics, Speech, and Signal Processing (ICASSP)

作者： Guohua Lv Baodong Zhang Yongbiao Gao Guixin Zhao Guotao Wang Juncan Wang Key Laboratory of Computing Power Network and Information Security Ministry of Education Shandong Computer Science Center (National Supercomputer Center in Jinan) Qilu University of Technology (Shandong Academy of Sciences) Jinan China Shandong Provincial Key Laboratory of Computing Power Internet and Service Computing Shandong Fundamental Research Center for Computer Science Jinan China Shandong Key Laboratory of Ubiquitous Intelligent Computing Jinan China Key Laboratory of New Generation Artificial Intelligence Technology and Its Interdisciplinary Application (Southeast University) Ministry of Education China Beihang University Beijing China

ISBN: (数字)9798350368741

ISBN: (纸本)9798350368758

Hyperspectral super-resolution involves combining low-resolution hyperspectral images with high-resolution multispectral images to produce a high-resolution hyperspectral image. Recently, although many methods for hyperspectral image super-resolution have been proposed, they often fail to fully utilize the high similarity among adjacent bands to enhance fusion performance. Therefore, we propose a grouping strategy-based progressive fusion network (GPFNet) for hyperspectral super-resolution. The core of GPFNet is the grouping strategy fusion block (GPF block), in which grouping-based spatial-spectral information fusion and spatial information refinement are performed. We design the spatial-spectral information fusion module (SSIFM) based on grouped convolutions to capture the feature differences from adjacent bands. To refine spatial details, we develop the spatial information enhancement module (SpaEM), which leverages the hierarchical features extracted by the multi-scale feature extraction module (MIEM). Additionally, a progressive fusion strategy, which involves using multiple upsampled hyperspectral images and downsampled multispectral images, further preserves spectral integrity and spatial details. Extensive experiments show that GPFNet outperforms state-of-the-art methods both qualitatively and quantitatively.

关键词： Convolution Superresolution Feature extraction Acoustics Data mining Speech processing Hyperspectral imaging

来源：评论

学校读者我要写书评

暂无评论

A Blockchain-Based Hierarchical Dynamic Collaborative Storage Model

A Blockchain-Based Hierarchical Dynamic Collaborative Storag...

引用

Smart World Congress (SWC), IEEE

作者： Qile Yuan Xiangzhi Liu Yu Wang Yunfeng Dong Haoyu Jia Xiaoming Wu Key Laboratory of Computing Power Network and Information Security Ministry of Education Shandong Computer Science Center (National Supercomputer Center in Jinan) Qilu University of Technology (Shandong Academy of Sciences) Jinan China Shandong Provincial Key Laboratory of Industrial Network and Information System Security Shandong Fundamental Research Center for Computer Science Jinan China

ISBN: (数字)9798331520861

ISBN: (纸本)9798331520878

The pervasive application of blockchain technology has led to a significant increase in data volume, presenting significant challenges to existing storage architectures in environments characterized by large volumes of data and highfrequency transactions. To enhance the scalability of blockchain, this study introduces an innovative framework, Hierarchical Dynamic Storage Management Chain Model, named HDSMC. The model consists of two stages. In the first phase of the model, when multimodal data files are uploaded, LBA techniques are employed to securely eliminate unnecessary redundancy, effectively preventing the storage of duplicate data. In the second phase, further processing is conducted on the generated blocks. This involves analyzing the frequency of block access to compute the variation in block “heat”. Based on the Block Heat (BH), the number of replicas for each block across participating nodes is then determined. Furthermore, the framework implements an adaptive tiering strategy based on data access frequency and its importance, which facilitates the timely migration of data to an appropriate storage level. A simulated Hyperledger Fabric environment was employed to deploy the HDSMC framework and to assess its performance against current storage methodologies. The empirical outcomes indicated that this approach could achieve lower storage expenditures while ensuring high reliability. This research not only presents a novel strategy for blockchain storage management but also suggests potential applications for storage optimization in other big data scenarios.

关键词： Heating systems Time-frequency analysis Sharding Scalability Storage management Redundancy Fabrics Blockchains Reliability Optimization

来源：评论

学校读者我要写书评

暂无评论

ID-YOLOv7: A Method for Steel Surface Defect Detection Combining Residual Attention and Dynamic Upsampling

ID-YOLOv7: A Method for Steel Surface Defect Detection Combi...

引用

Smart World Congress (SWC), IEEE

作者： Shanshan Liu Xiangzhi Liu Jiazheng Man Bingyang Cao Yunfeng Dong Xiaoming Wu Key Laboratory of Computing Power Network and Information Security Ministry of Education Shandong Computer Science Center (National Supercomputer Center in Jinan) Qilu University of Technology (Shandong Academy of Sciences) Jinan China Shandong Provincial Key Laboratory of Industrial Network and Information System Security Shandong Fundamental Research Center for Computer Science Jinan China

ISBN: (数字)9798331520861

ISBN: (纸本)9798331520878

Industrial defect detection is a crucial challenge in computer vision. The accuracy of detecting defects on steel surfaces is hindered by spatially varying noise. To address this challenge, we propose an efficient method for steel surface defect detection. Firstly, considering the impact of the shape and proportions of bounding boxes on localization accuracy, we developed an Adaptive Shape-Inner Intersection over Union (ASIIoU) loss function. This function dynamically adjusts auxiliary bounding boxes and takes into account their shape and scale to optimize the loss, thereby improving the accuracy of bounding box regression loss. Secondly, to enhance the ability of the model to learn local features, we designed a Content-Guided Residual Attention Block (CGRAB). It captures rich local features across different convolutional layers via skip connections while discarding irrelevant information. Lastly, to avoid the potential issue of detail blurring caused by traditional upsampling methods, we optimized the upsampling process to preserve key details and structural information in the image, reducing unnecessary noise amplification. Experimental validation demonstrates the effectiveness of our method, which achieved a $76.7 \% \mathrm{mAP}$ on the NEU-DET dataset.

关键词： Location awareness Accuracy Shape Noise Focusing Object detection Feature extraction Steel Surface treatment Defect detection

来源：评论

学校读者我要写书评

暂无评论

Research and Application of Speech Transmission Index in Acoustic Leakage Detection

Research and Application of Speech Transmission Index in Aco...

引用

International Congress on Image and Signal Processing, BioMedical Engineering and Informatics (CISP-BMEI)

作者： Qiuyun Hao FuQiang Wang Peng Zhang XiaoFeng Ma Jianqiang Zhang Wei Zhao Key Laboratory of Computing Power Network and Information Security Ministry of Education Shandong Computer Science Center (National Supercomputer Center in Jinan) Qilu University of Technology (Shandong Academy of Sciences) Jinan China Shandong Provincial Key Laboratory of Industrial Network and Information System Security Shandong Fundamental Research Center for Computer Science Jinan China

ISBN: (数字)9798331507398

ISBN: (纸本)9798331507404

Acoustic information in confidential premise can be leaked through channels such as walls, pipes, doors and windows. Acoustic leakage level needs to be measured through specific indicators. Acoustic leakage evaluation indicator includes subjective evaluation indicator and objective evaluation indicator. The subjective evaluation indicator is a qualitative evaluation of acoustic leakage through subjective auditory perception of human ear, which can directly reflect the degree of acoustic leakage in confidential premise. At present, the objective evaluation indicator is mainly sound transmission loss (STL) to reflect objective degree of acoustic leakage in confidential premise. Therefore, it is relatively singular. Through study of Speech Transmission Index (STI), speech transmission index for public address systems (STIPA), which is a simplified form of FULL STI, is applied to acoustic leakage detection in this paper. STI has been proposed as an objective evaluation indicator and applied to acoustic leakage detection and acoustic information security. In order to confirm the effectiveness of objective evaluation indicator, a large number of related experimental studies have been conducted in the paper. Through experiments, it has been proven that STI can objectively evaluate level of acoustic leakage and level of acoustic information security in confidential premise. STI can better predict level of Chinese clarity under various complex transmission conditions. It provides a reliable basis for speech privacy of confidential premise and evaluation of architectural acoustic, sound reinforcement system and voice communication system.

关键词： Privacy Communication systems information security Ear Signal processing Propagation losses Acoustics Indexes Reliability Informatics

来源：评论

学校读者我要写书评

暂无评论

Reputation based Adaptive Federated Learning in Medical Scenarios

Reputation based Adaptive Federated Learning in Medical Scen...

引用

IEEE International Conference on Data Science in Cyberspace (DSC)

作者： Lianhai Wang Tianrui Liu Yingxiaochun Wang Xinlei Wang Qi Li Key Laboratory of Computing Power Network and Information Security Ministry of Education Shandong Computer Science Center (National Supercomputer Center in Jinan) Qilu University of Technology (Shandong Academy of Sciences) Jinan China Shandong Provincial Key Laboratory of Industrial Network and Information System Security Shandong Fundamental Research Center for Computer Science Jinan China

ISBN: (数字)9798350391367

ISBN: (纸本)9798350391374

Federated Learning (FL) is a distributed machine learning paradigm designed to address data silos and protect data privacy. However, in medical scenarios, the heterogeneity in data quality and the non-independent and identically distributed (Non-IID) nature of medical data from different hospitals can lead to degraded model performance in FL. To tackle this challenge, this paper proposes an adaptive federated learning framework based on reputation. By using a reputation evaluation model, the framework adaptively adjusts the aggregation weights of the models to enhance the performance of the global model. Additionally, our reputation evaluation model, integrated with blockchain technology, can rapidly identify malicious clients and effectively prevent poisoning attacks. Experimental results on two medical datasets demonstrate that this framework achieves higher model accuracy and faster convergence.

关键词： Adaptation models Data privacy Federated learning Hospitals Data integrity Computational modeling Evaluation models Data models Security Convergence

来源：评论

学校读者我要写书评

暂无评论

Self-Evaluation of Large Language Model based on Glass-box Features

arXiv

引用

arXiv 2024年

作者： Huang, Hui Qu, Yingqi Liu, Jing Yang, Muyun Xu, Bing Zhao, Tiejun Lu, Wenpeng Faculty of Computing Harbin Institute of Technology Harbin China Baidu Inc. Beijing China Key Laboratory of Computing Power Network and Information Security Ministry of Education Shandong Computer Science Center Qilu University of Technology Jinan China

The proliferation of open-source Large Language Models (LLMs) underscores the pressing need for evaluation methods. Existing works primarily rely on external evaluators, focusing on training and prompting strategies. However, a crucial aspect – model-aware glass-box features – is overlooked. In this study, we explore the utility of glass-box features under the scenario of self-evaluation, namely applying an LLM to evaluate its own output. We investigate various glass-box feature groups and discovered that the softmax distribution serves as a reliable quality indicator for self-evaluation. Experimental results on public benchmarks validate the feasibility of self-evaluation of LLMs using glass-box features1 Copyright © 2024, The Authors. All rights reserved.

关键词： Open systems

来源：评论

学校读者我要写书评

暂无评论

SGNet: A Multi-Scale Method for Steel Surface Defect Detection

SGNet: A Multi-Scale Method for Steel Surface Defect Detecti...

引用

Smart World Congress (SWC), IEEE

ISBN: (数字)9798331520861

ISBN: (纸本)9798331520878

The large variation in target scales within steel surface defect images often results in low detection accuracy. To address this issue, we have developed a multi-scale defect detection method named SGNet. Initially, we introduced a Self-Guided Bi-Level Routing Transformer (SG-BiFormer) as the backbone network of our model. SG-BiFormer utilizes a self-guided strategy to model global information and dynamically reallocates tokens based on the importance of image regions, increasing token quantity in significant areas to preserve fine-grained details and reducing it in less critical areas. This strategy significantly enhances the accuracy and efficiency of the model in handling multi-scale targets. Subsequently, we incorporated Grouped Separable Convolutions (GSConv) to manage local information of multi-scale features. GSConv automatically adjusts the size and shape of the convolution kernels according to different input features. This design not only improves global information processing capabilities but also ensures sensitive and precise detail capture. Lastly, we employed MPDIoU as the loss function for our model, which optimizes corner positioning to tackle the challenges of regression tasks, substantially improving the accuracy of the model in handling complex defects. Experimental evidence confirms the effectiveness of this method. Our SGNet achieved a 77.1% mAP on the NEU-DET dataset, demonstrating outstanding performance.

关键词： Visualization Accuracy Shape Detectors Transformers Routing Feature extraction Steel Kernel Defect detection

来源：评论

学校读者我要写书评

暂无评论

Infringement Detection and Traceability Scheme Based on Ciphertext Feature Extraction

Infringement Detection and Traceability Scheme Based on Ciph...

引用

IEEE International Conference on Data Science in Cyberspace (DSC)

作者： Lianhai Wang Xinlei Wang Tianrui Liu Yingxiaochun Wang Qi Li Key Laboratory of Computing Power Network and Information Security Ministry of Education Shandong Computer Science Center (National Supercomputer Center in Jinan) Qilu University of Technology (Shandong Academy of Sciences) Jinan China Shandong Provincial Key Laboratory of Industrial Network and Information System Security Shandong Fundamental Research Center for Computer Science Jinan China

ISBN: (数字)9798350391367

ISBN: (纸本)9798350391374

The rapid growth of healthcare data transactions has fostered win-win developments in healthcare technology innovation and economic benefits. However, the issue of secondary transactions makes data ownership vulnerable to infringement due to the easily replicable nature of the data. As the importance of ownership and privacy increases, more and more users protect the privacy of data through encryption methods, making it impossible for existing methods to make infringement judgments on ciphertext data. To solve this problem, we propose an infringement detection and traceability scheme based on feature extraction of ciphertext data. Firstly, the characteristics of fully homomorphic encryption are utilized to obtain data features in the ciphertext state, and the plaintext state of these features is used as the basis for infringement judgment. Additionally, during the data transaction process, the ownership registration of the sales data is verified through smart contracts to ensure the effective association between the transaction data and the seller’s address. The experimental results show that the scheme can accurately determine infringement while protecting data security and privacy.

关键词： Economics Data privacy Technological innovation Data security Smart contracts Cyberspace Medical services Data science Feature extraction Homomorphic encryption

来源：评论

学校读者我要写书评

暂无评论

Workflow Scheduling Method in Multi-Cluster Environments Using Tasks' Spatiotemporal Attributes

Workflow Scheduling Method in Multi-Cluster Environments Usi...

引用

Electronic information Engineering and Computer Communication (EIECC), International Conference on

作者： Yudong Geng Jing Chen Honghua Yan Qingshui Hao Aimin Zhang Tianxiang Zhang Mingchao Ji Tonglin Fu Key Laboratory of Computing Power Network and Information Security Ministry of Education Shandong Computer Science Center (National Supercomputer Center in Jinan) Qilu University of Technology (Shandong Academy of Sciences) Jinan China Shandong Provincial Key Laboratory of Computing Power Internet and Service Computing Shandong Fundamental Research Center for Computer Science Jinan China Shandong Greenthink Electric Power Technology Co. Ltd Jinan China

ISBN: (数字)9798331534622

ISBN: (纸本)9798331534639

Nowadays, computing resources are widely distributed and diverse, and the demand for efficient execution of various workflow tasks on wide-area resources has been raised. multi-objection scheduling method based on space-time attribute of tasks. A scheduling scheme is generated through the three stages of workflow scheduling queue generation, resource selection combined with task spatiotemporal attributes and task and resource matching, and the selected resources of workflow are run according to this scheduling scheme. This method constructs a scheduling problem model to minimize workflow completion time, cost and resource matching degree, and finds the optimal scheduling scheme to improve task execution efficiency. In the process of the experiment, the workflow is actually run in the multi-cluster environment, and the experiment proves that it has a good effect in reducing the completion time and cost of workflow.

关键词： Sequential analysis Costs Processor scheduling Filtering Optimal scheduling Scheduling Spatiotemporal phenomena

来源：评论

学校读者我要写书评

暂无评论

Continuous Verification of Catastrophic Recalling in Machine Unlearning via Adversarial Testing 9

Continuous Verification of Catastrophic Recalling in Machine...

引用

9th IEEE International Conference on Data Science in Cyberspace, DSC 2024

作者： Qu, Y. Neil Ji, Zonghao Cui, Lei Zhang, Chong Liu, Lei Tian, Zhiyi Qilu University of Technology Shandong Academy of Sciences Key Laboratory of Computing Power Network and Information Security Ministry of Education Shandong Computer Science Center Jinan250031 China Shandong Fundamental Research Center for Computer Science Shandong Provincial Key Laboratory of Computer Networks Jinan250031 China Deakin University School of IT VIC3125 Australia Xidian University Guangzhou Institute of Technology Guangzhou510555 China University of Technology School of Computer Science SydneyNSW2007 Australia

ISBN: (纸本)9798350391367

Throughout the process of machine unlearning, in order to fulfill the requirements of machine unlearning and to comply with the rules of the right to be forgotten, it is natural to verify whether the unlearning model actually removes data points. Therefore, machine unlearning verification plays a crucial role. However, existing validation methods only perform one validation, and it is still challenging to determine whether subsequent updates of the model still maintain the unlearning effect. To address this challenge, we propose a novel continuous verification of catastrophic recalling in machine unlearning framework that aims to continuously evaluate whether a machine learning model effectively forgets specified data after an update. In this work, we first introduce adversarial testing as a crucial part of the overall framework. Verification then continues as the unlearned model is updated (adding or continuing to forget data points). We also analyzed the possibility that learning new data after unlearning may diminish the effect. Overall, our continuous unlearning verification framework provides a proactive approach to protecting individual privacy in the machine learning domain, thus facilitating the generation of more robust unlearning algorithms. © 2024 IEEE.

关键词： Adversarial machine 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：