检索结果-内蒙古大学图书馆

arXiv 2024年

作者： Li, Kaiyu Cao, Xiangyong Deng, Yupeng Song, Jiayi Liu, Junmin Meng, Deyu Wang, Zhi School of Software Engineering Xi’an Jiaotong University Xi’an710049 China School of Computer Science and Technology Ministry of Education Key Lab For Intelligent Networks and Network Security Xi’an Jiaotong University Xi’an710049 China Aerospace Information Research Institute Chinese Academy of Sciences Beijing100094 China School of Mathematics and Statistics Ministry of Education Key Lab of Intelligent Networks and Network Security Xi’an Jiaotong University Shaanxi Xi’an China Guangdong Guangzhou China

Change Detection (CD) aims to identify pixels with semantic changes between images. However, annotating massive numbers of pixel-level images is labor-intensive and costly, especially for multi-temporal images, which require pixel-wise comparisons by human experts. Considering the excellent performance of visual language models (VLMs) for zero-shot, open-vocabulary, etc. with prompt-based reasoning, it is promising to utilize VLMs to make better CD under limited labeled data. In this paper, we propose a VLM guidance-based semisupervised CD method, namely SemiCD-VL. The insight of SemiCD-VL is to synthesize free change labels using VLMs to provide additional supervision signals for unlabeled data. However, almost all current VLMs are designed for single-temporal images and cannot be directly applied to bi- or multi-temporal images. Motivated by this, we first propose a VLM-based mixed change event generation (CEG) strategy to yield pseudo labels for unlabeled CD data. Since the additional supervised signals provided by these VLM-driven pseudo labels may conflict with the original pseudo labels from the consistency regularization paradigm (e.g. FixMatch), we propose the dual projection head for de-entangling different signal sources. Further, we explicitly decouple the bi-temporal images semantic representation through two auxiliary segmentation decoders, which are also guided by VLM. Finally, to make the model more adequately capture change representations, we introduce contrastive consistency regularization by constructing feature-level contrastive loss in auxiliary branches. Extensive experiments show the advantage of SemiCD-VL. For instance, SemiCD-VL improves the FixMatch baseline by +5.3 IoUc on WHU-CD and by +2.4 IoUc on LEVIR-CD with 5% labels, and SemiCD-VL requires only 5% to 10% of the labels to achieve performance similar to the supervised methods. In addition, our CEG strategy, in an un-supervised manner, can achieve performance far superior to state-of-t

关键词： Visual languages

来源：评论

学校读者我要写书评

暂无评论

GNN-ASG: A Double Feature Selection-based Adversarial Sample Generation Method in Industrial Control System

GNN-ASG: A Double Feature Selection-based Adversarial Sample...

引用

International Conference on computer Supported Cooperative Work in Design

作者： Lijuan Xu Zhiang Yao Dawei Zhao Xin Li Ministry of Education Shandong Computer Science Center (National Supercomputer Center in Jinan) Key Laboratory of Computing Power Network and Information Security Qilu University of Technology (Shandong Academy of Sciences) Jinan China Shandong Provincial Key Laboratory of Computer Networks Shandong Fundamental Research Center for Computer Science Jinan China

ISBN: (数字)9798350349184

ISBN: (纸本)9798350349191

Due to the unique constraints of industrial control data, industrial control adversarial sample attacks are particularly challenging. Existing methods strive to conduct adversarial sample attacks under the conditions of satisfying industrial control data constraints, however, the results are not ideal. Therefore, this study proposes a new adversarial sample generation method GNN-ASG based on double feature selection. GNN-ASG uses data constraints to ensure the rationality of generated data, and uses Graph Deviation network (GDN) and Autoencoder to improve the quality and versatility of adversarial samples. A new adversarial sample evaluation metrics Adversarial Sample Attack Impact Rate (ASAIR) is proposed to address the problem that existing evaluation metrics are difficult to accurately judge the effectiveness of adversarial sample attacks. This method considers the principle and application environment of adversarial samples, and fully demonstrates the practical effect of adversarial samples. In a comprehensive experiment conducted on three public datasets, GNN-ASG achieves an impressive ASAIR of 21.83%, higher than existing methods of 13.94%. This paper demonstrates the versatility and effectiveness of GNN-ASG by comparing its performance with three state-of-the-art adversarial sample generation methods on four anomaly detection models. GNN-ASG can maximally reduce the F1-Score of the detection model by 0.7605.

关键词： Measurement Federated learning Industrial control Feature extraction Anomaly detection Glass box

来源：评论

学校读者我要写书评

暂无评论

A Novel Collaborative Edge Caching via User Preference Awareness and Adaptive Clustering in MENs

A Novel Collaborative Edge Caching via User Preference Aware...

引用

IEEE Internatonal Conference on Mobile Adhoc and Sensor Systems (MASS)

作者： Pengfei Gu Huiling Shi Hao Hao Mingfa Li Wei Zhang 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 Computer Networks Shandong Fundamental Research Center for Computer Science Jinan China

ISBN: (数字)9798350363999

ISBN: (纸本)9798350364002

In this paper, we propose a collaborative caching placement scheme leveraging user preference awareness and adaptive clustering (UPAAC), which jointly considers the heterogeneity of user preferences and user activity to reduce user access latency and improve the quality of service. Specifically, we first utilize Bidirectional Long Short-Term Memory (Bi-LSTM) to capture dynamic user behavior to predict user preferences, and then density-based spatial clustering of applications with noise (DBSCAN) is used to cluster users based on the predicted user preferences for more efficient cache placement. Furthermore, the cache placement optimal problem is defined as the latency reduction maximization problem under the constraint of the cache capacity and latency threshold, and a genetic algorithm (GA) is used to optimize the problem and design the cache placement strategy. The simulation outcomes indicate that the suggested approach outperforms the conventional caching methods by significantly decreasing content access latency and enhancing the cache hit ratio.

关键词： Noise Collaboration Quality of service Prediction algorithms Smart systems Probability distribution Timing History Long short term memory Genetic algorithms

来源：评论

学校读者我要写书评

暂无评论

An Efficient Token Mixer Model for Sheet Metal Defect Detection

An Efficient Token Mixer Model for Sheet Metal Defect Detect...

引用

IEEE International Conference on Systems, Man and Cybernetics

作者： Huijuan Hao Sijian Zhu Changle Yi Yu Chen Hongge Zhao Yue Feng Rong Yang 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 Computer Networks Shandong Fundamental Research Center for Computer Science Jinan China

ISBN: (数字)9781665410205

ISBN: (纸本)9781665410212

Defects such as scratches, patches, and cracks frequently occur during sheet metal production. However, the low detection accuracy and slow processing speed of industrial defect detection models significantly impede enterprise production efficiency. The aforementioned issues primarily manifest in three aspects. Firstly, the model complexity and computational overhead are substantial. Secondly, detecting small local defects poses a significant challenge. Thirdly, extracting global features, such as elongated scratches, proves to be difficult. To address these challenges, this paper introduces a novel network architecture called SATRNet. Firstly, within the model backbone, the STR module is devised. Through incorporation of the sparse self-attention method and the CNNs parallel vision Transformer model in the shallow layers, this module significantly enhances the model's capability to extract global features. Secondly, the SCATR module is designed in this paper. By substituting self-attention with the designed SCA soft attention as the token mixer, the module aims to enhance detection accuracy while reducing the number of parameters, thereby fundamentally addressing the problem of model complexity. Finally, this paper presents the GCD bottleneck convolution module. This module combines shallow and deep features, enabling the fusion of more information beneficial for detection, thereby achieving improved efficacy in capturing minute targets. Experiments demonstrate that SATRNet surpasses existing advanced models in detection accuracy on public datasets.

关键词： Accuracy Computational modeling Metals Production network architecture Feature extraction Transformers Complexity theory Mixers Defect detection

来源：评论

学校读者我要写书评

暂无评论

Lightweight Malicious Encrypted Traffic Detection Method Based on KL Image Compression

Lightweight Malicious Encrypted Traffic Detection Method Bas...

引用

IEEE International Conference on Data Science in Cyberspace (DSC)

作者： Yuecheng Wen Xiaotong Li Haiqing Lv Menglu Wang Wenbo Zuo 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 Computer Networks Shandong Fundamental Research Center for Computer Science Jinan China

ISBN: (数字)9798350391367

ISBN: (纸本)9798350391374

Timely and accurate detection of and response to potentially malicious encrypted traffic is crucial in cybersecurity strategies. Convolutional neural network-based approaches that convert traffic loads into images for analysis have made significant progress in malicious encrypted traffic detection, but they suffer from the challenges of large number of participants, large network size, and high demand for computational resources, especially when edge computing devices are limited. Therefore, there is an urgent need for a lightweight, efficient and accurate malicious encrypted traffic detection model. This paper proposes a malicious encrypted traffic detection model whose core idea is to utilize KL image compression technique. The model first converts the raw traffic data into grayscale images and applies KL transform to compress the images efficiently. Subsequently, a convolutional neural network model with lightweight design is used to classify the compressed image for fast and accurate malicious encrypted traffic detection. After validating on both the public traffic dataset and the lab’s self-collected dataset, the results indicate that our method requires just 0.0216 seconds to process a single input sample, achieving an average classification accuracy of 98%. These findings highlight its practicality and excellence in addressing the real-world challenges of malicious encrypted traffic.

关键词： Image coding Accuracy Computational modeling Transforms Telecommunication traffic Gray-scale Data models Mobile handsets Cryptography Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

A Patent Fast Recommendation Algorithm Based on User Requirements Combined with an Inverted Table

A Patent Fast Recommendation Algorithm Based on User Require...

引用

Advanced information, Mechanical Engineering, Robotics and Automation (AIMERA), IEEE International Conference on

作者： Wan Zheng Qingyan Ding Na Li Yu Pan Xuecheng Dong 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 Computer Networks Shandong Fundamental Research Center for Computer Science Jinan China

ISBN: (数字)9798350343335

ISBN: (纸本)9798350343342

Patent recommendation has always been an essential service for the innovation and development of all walks of life. The matching of demand and patent information can enable enterprises to quickly obtain the innovation and development trend of the industry. The current patent recommendation method has the problem of slow recommendation speed and insufficient comprehensive recommendation results. According to the existing problems, a patent fast recommendation algorithm based on user needs combined with inverted table is proposed. The algorithm using dual track recommended way for accurate recommendation and extensive recommendation, in the process of recommendation built by patent data inversion table and document list auxiliary recommendation to improve recommendation speed and accuracy, when accurate recommendation when using improved requirements based on GPT-2 compression model recommended after demand text compression, when widely recommended using TF-IDF mechanism for demand text of each paragraph after keyword extraction, finally generate widely recommendation list and accurate recommendation list. This recommendation method can not only realize the fast and accurate recommendation of patents, but also provide users with more innovative ideas for extensive recommendation. The experiments show that the proposed recommendation algorithm has higher speed and accuracy than the original recommendation algorithm.

关键词： Patents Vocabulary Technological innovation Accuracy Semantics Predictive models Writing Prediction algorithms Data mining Robotics and automation

来源：评论

学校读者我要写书评

暂无评论

Improvement in the Transferability of Target Adversarial Examples Based on Data Augmentation

Improvement in the Transferability of Target Adversarial Exa...

引用

IEEE International Conference on Systems, Man and Cybernetics

作者： Kun Wang Xing Jin Fuqiang 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 Computer Networks Shandong Fundamental Research Center for Computer Science Jinan China

ISBN: (数字)9781665410205

ISBN: (纸本)9781665410212

Despite the high performance of deep neural networks in image classification, they are very sensitive to those adversarial examples with small perturbations, which is likely to cause classification errors. Currently, there are various adversarial attack methods developed, but the adversarial samples generated using most methods are still disadvantaged by the poor target ransferability of cross-model in black-box scenarios. To address this issue, a new method called Adaptive Variation Method (AVM) is proposed in this paper to improve the transferability of targeted adversarial examples in black-box scenarios. AVM relies on Adjustment of diversified inputs(ADI) method to achieve data enhancement on the input image and Maximum absolute difference (MD) method to calculate the loss for adversarial perturbation. Specifically, the Resized-Diverse-Inputs (RDI) method is improved in this paper to propose a new diversified input method, namely the Adjustment of diversified inputs (ADI) method. By means of repeated scaling and random filling performed on the input image, the image becomes relatively rough, with its input diversity and random flexibility improved. In addition, the Maximum absolute difference (MD) method is proposed in this paper to enhance the advantage of adversarial samples by minimizing the maximum absolute difference between the adversarially perturbed global image and local image. According to the experimental results, this method improves the success rate in the presence of transfer-based targeted attacks.

关键词： Deep learning Perturbation methods Data enhancement Closed box Data augmentation Robustness Filling Cybernetics Overfitting Image classification

来源：评论

学校读者我要写书评

暂无评论

CB-YOLO: A Small Object Detection Algorithm for Industrial Scenarios

CB-YOLO: A Small Object Detection Algorithm for Industrial S...

引用

IEEE International Conference on Systems, Man and Cybernetics

作者： Hao Chen Xiaoming Wu Zhanzhi Su Ying Zhao Yunfeng Dong Xiangzhi Liu 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 Computer Networks Shandong Fundamental Research Center for Computer Science Jinan China

ISBN: (数字)9781665410205

ISBN: (纸本)9781665410212

In certain specific industrial scenarios, smoking and cellphone usage are strictly prohibited behaviors. In these scenarios, it is crucial to rapidly and accurately detect smoking and cell phone usage, and promptly issue warnings, to ensure industrial safety. The detection of prohibited behaviors using computer vision has gained attention from researchers with the development of artificial intelligence. However, detecting small objects like cigarettes and cell phones in complex backgrounds and varying angles poses a challenge due to their changing shapes. In this paper, we propose a detection model called CB-YOLO, which utilizes YOLOv7 as the baseline model, for detecting smoking and mobile phone usage behavior. Firstly, we propose a new channel space pyramid network called CSPPF that pools features at different scales and introduces scale focusing on improving the perceptual ability of the network to better handle targets at various scales and locations. Secondly, we propose an enhanced feature pyramid called AWBFPN. This introduces additional learnable weight parameters to improve the model's ability to fuse multi-scale features effectively. Finally, we have also proposed a new loss function called Size-IoU. The experimental results demonstrate that our algorithm outperforms the baseline model. Specifically, it achieves a 3.9% improvement in Precision, a 2.1% improvement in Recall, a 6.4% improvement in mAP@0.5, and a 2% improvement in mAP@0.5:0.95 on the Phone_check dataset. Similarly, on the Smoking dataset, our algorithm achieves a 3.3% improvement in Precision, a 1.9% improvement in Recall, a 1.7% improvement in mAP@0.5, and a 0.3% improvement in mAP@0.5:0.95.

关键词： Fuses Shape Cellular phones Focusing Object detection Robustness Mobile handsets Safety Iterative methods Cybernetics

来源：评论

学校读者我要写书评

暂无评论

OCF-HP: An Offline Compaction Framework Based on Hotspot Prediction for Data Lakehouse

OCF-HP: An Offline Compaction Framework Based on Hotspot Pre...

引用

Electronic Technology and information Science (ICETIS), International Conference on

作者： Bowen Sun Xiaohan Yang Ying Guo Zhihao Zhao Jing Chen Hu Zhang Jibin 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 Computer Networks Shandong Fundamental Research Center for Computer Science Jinan China

ISBN: (数字)9798350388343

ISBN: (纸本)9798350388350

With the rapid advancement of information technologies, the global data volume is increasing at an unprecedented speed. Data Lakehouse has become an effective solution for real-time aggregation of multimodal heterogeneous data and supporting multiple query modes. It is worth noting that due to the column-based storage commonly employed in Data Lakehouse and the resulting mechanisms such as Copy-on-Write and Merge-On-Read, data query speed, data freshness, and storage costs have become huge challenges. In this paper, an Offline Compaction Framework based on Hotspot Prediction (OCF-HP) was designed to address the above challenges. Specifically, we analyzed the Timeline data in Hudi which is the mainstream Data Lakehouse framework, then predicted the hotspot periods of data writing based on the improved LSTM model using such data, and realized automatic trigger of Hudi offline compaction tasks in reasonable non-hot periods. The experiment results show that OCF-HP provides Hudi with a more optimized offline compaction solution so that data lakehouse managers could achieve better query speed and data freshness with reducing storage costs in Hudi's read optimized mode.

关键词： Analytical models Costs Data integrity Predictive models Writing Data models Compaction

来源：评论

学校读者我要写书评

暂无评论

Patent Recommendation Methods Based on Transformer Encoders and Regularization Strategies

Patent Recommendation Methods Based on Transformer Encoders ...

引用

Advanced information, Mechanical Engineering, Robotics and Automation (AIMERA), IEEE International Conference on

ISBN: (数字)9798350343335

ISBN: (纸本)9798350343342

Patent recommendation service has become a major demand of all walks of life in the society. Patent data collection and analysis can not only help enterprises grasp the development status of the industry, but also provide decision support for them in the research and development process of science and technology services. In view of the problem of inaccurate patent recommendation caused by the inconsistency between the search text and the patent text length and the low accuracy of traditional patent recommendation, a word and sentence dual processing model based on Transformer encoder and regularization strategy is proposed to complete the patent recommendation task from the level of sentence granularity and word granularity. The sentence granularity feature extraction module uses the transformer encoder superimposed multi-head attention method to obtain the feature information between the sentence vector groups, so that the sentence vector fuses the feature information of each sentence, and solves the problem of low recommendation accuracy caused by the inconsistent length of the patent text and the search text; The word granularity feature extraction module uses the twin network model to extract the feature information of the sentence granularity vector using the Transformer encoder using the double splicing operation, improve the model performance and model accuracy; Introduce a strategy based on Dropout and symmetric JS divergence regularization, improve the model generalization ability and accuracy. Experiments have proved that it outperforms SBERT, BERT, MGMSN and SVM models in terms of performance.

关键词： Patents Accuracy Fuses Splicing Feature extraction Transformers Search problems Vectors Data models Data mining

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：