检索结果-内蒙古大学图书馆

IEEE International Conference on Data Science in Cyberspace (DSC)

作者： Ning Su Lijuan Sun Yutong Gao Jingchen Wu Xu Wu Key Laboratory of Trustworthy Distributed Computing and Service Ministry of Education School of Cyberspace Security BUPT School of Economics and Management Beijing University of Posts and Telecommunications Beijing China Beijing Key Lab of Traffic Data Analysis and Mining Beijing Jiaotong University Beijing China School of Computer Science (National Pilot Software Engineering School) BUPT

Video violence detection aims to locate the time window in which violent behavior occurs. Most methods focus on utilizing RGB features directly or only fusing RGB and audio features, ignoring the effective exploitation of motion information carried in optical flow. This lack of emphasis on motion information may impact the overall accuracy of violence detection. Moreover, we observe that videos contain strong local correlations, so it is insufficient to analyze only from a holistic perspective without capturing finer details. Therefore, in this paper, we design a novel Global-and-Local Cross-Modal Network (GL-CMN) for violence detection, which effectively integrates motion information and multi-granularity features from target videos. Specifically, we first propose a Motion-Guided Attention Module (MGAM) to obtain enhanced visual features by calibrating RGB features through optical flow features. Secondly, The enhanced features are simultaneously fed into two parallel branches of the network. The global branch fuses the visual and audio features into holistic representations. The local branch extracts multi-scale temporal dependencies through dilated convolutions. Experiments demonstrate that our method exhibits significant improvement compared to previous state-of-the-art methods on the XD-Violence dataset.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Excitement Surfeited Turns to Errors: Deep Learning Testing Framework Based on Excitable Neurons

arXiv

引用

arXiv 2022年

作者： Jin, Haibo Chen, Ruoxi Zheng, Haibin Chen, Jinyin Cheng, Yao Yu, Yue Liu, Xianglong College of Information Engineering Zhejiang University of Technology Hangzhou China Institute of Cyberspace Security Zhejiang University of Technology Hangzhou China Huawei International Singapore National Laboratory for Parallel and Distributed Processing College of Computer National University of Defense Technology Changsha China State Key Laboratory of Software Development Environment Beihang University Beijing China

Despite impressive capabilities and outstanding performance, deep neural networks (DNNs) have captured increasing public concern about their security problems, due to their frequently occurred erroneous behaviors. Therefore, it is necessary to conduct a systematical testing for DNNs before they are deployed to real-world applications. Existing testing methods have provided fine-grained metrics based on neuron coverage and proposed various approaches to improve such metrics. However, it has been gradually realized that a higher neuron coverage does not necessarily represent better capabilities in identifying defects that lead to errors. Besides, coverage-guided methods cannot hunt errors due to faulty training procedure. So the robustness improvement of DNNs via retraining by these testing examples are unsatisfactory. To address this challenge, we introduce the concept of excitable neurons based on Shapley value and design a novel white-box testing framework for DNNs, namely DeepSensor. It is motivated by our observation that neurons with larger responsibility towards model loss changes due to small perturbations are more likely related to incorrect corner cases due to potential defects. By maximizing the number of excitable neurons concerning various wrong behaviors of models, DeepSensor can generate testing examples that effectively trigger more errors due to adversarial inputs, polluted data and incomplete training. Extensive experiments implemented on both image classification models and speaker recognition models have demonstrated the superiority of DeepSensor. Compared with the state-of-the-art testing approaches, DeepSensor can find more test errors due to adversarial inputs (∼ ×1.2), polluted data (∼ ×5) and incompletely-trained DNNs (∼ ×1.3). Additionally, it can help DNNs build larger l2-norm robustness bound (∼ ×3) via retraining according to CLEVER's certification. We further provide interpretable proofs for effectiveness of DeepSensor via excitable neuro

关键词： Deep neural networks

来源：评论

学校读者我要写书评

暂无评论

Ontology-based Knowledge Modeling for Using Physical Effects

引用

Procedia Engineering 2015年 131卷 601-615页

作者： Wei Yan Cecilia Zanni-Merk François Rousselot Denis Cavallucci Pierre Collet School of Information Science and Engineering Shandong Normal University Jinan 250014 China Shandong Provincial Key Laboratory for Novel Distributed Computer Software Technology Jinan 250014 China LGECO / INSA Strasbourg 24 Boulevard de la Victoire 67084 Strasbourg Cedex France ICube / BFO Team (UMR CNRS 7005) – Pôle API BP 10413 67412 Illkirch Cedex France

Su-Field analysis, as one of the inventive problem solving tools, can be used to analyse and improve the efficacy of the technical system. Generally, the process of using Su-Field model to solve a specific inventive problem includes: building a Problem Model, mapping to a Generic Problem Model, finding a Generic Solution Model based on the corresponding inventive standard, and finally establishing and instantiating a Solution Model. As one of the most important phases of Su-Field analysis, the last step is normally implemented manually with the help of physical effects, which link generic technical functions with specific applications and systems. The physical effects compatible with the context of the specific problem should be chosen to assist the users to instantiate the Solution Model. However, the physical effects and the specific problems are built at different levels of abstraction, and it is difficult for the users to choose, that is, given a certain function, too many physical effects are chosen while with the detailed context of the problem, no physical effect is returned. This paper firstly proposes a new way of representing physical effects using the change of two states, that is, the couple of two states before and after applying physical effects. Then, the knowledge about using physical effects is formalized in OWL (Ontology Web Language) - an ontology language for semantic web, and the constraint knowledge, such as the condition to use each kind of physical effect, is formalized in SWRL (Semantic Web Rule Language) - a rule language. Finally, the reasoning process of using physical effects is performed with the support of JESS (Java Expert System Shell) rule engine.

关键词： Physical Effects Ontology OWL (Ontology Web Language) SWRL (Semantic Web Rule Language) JESS (Java Expert System Shell)

来源：评论

学校读者我要写书评

暂无评论

DarkSAM: Fooling Segment Anything Model to Segment Nothing 38

DarkSAM: Fooling Segment Anything Model to Segment Nothing

引用

38th Conference on Neural Information Processing Systems, NeurIPS 2024

作者： Zhou, Ziqi Song, Yufei Li, Minghui Hu, Shengshan Wang, Xianlong Zhang, Leo Yu Yao, Dezhong Jin, Hai National Engineering Research Center for Big Data Technology and System China Services Computing Technology and System Lab China Cluster and Grid Computing Lab China Hubei Engineering Research Center on Big Data Security China Hubei Key Laboratory of Distributed System Security China School of Computer Science and Technology Huazhong University of Science and Technology China School of Cyber Science and Engineering Huazhong University of Science and Technology China School of Software Engineering Huazhong University of Science and Technology China School of Information and Communication Technology Griffith University Australia

Segment Anything Model (SAM) has recently gained much attention for its outstanding generalization to unseen data and tasks. Despite its promising prospect, the vulnerabilities of SAM, especially to universal adversarial perturbation (UAP) have not been thoroughly investigated yet. In this paper, we propose DarkSAM, the first prompt-free universal attack framework against SAM, including a semantic decoupling-based spatial attack and a texture distortion-based frequency attack. We first divide the output of SAM into foreground and background. Then, we design a shadow target strategy to obtain the semantic blueprint of the image as the attack target. DarkSAM is dedicated to fooling SAM by extracting and destroying crucial object features from images in both spatial and frequency domains. In the spatial domain, we disrupt the semantics of both the foreground and background in the image to confuse SAM. In the frequency domain, we further enhance the attack effectiveness by distorting the high-frequency components (i.e., texture information) of the image. Consequently, with a single UAP, DarkSAM renders SAM incapable of segmenting objects across diverse images with varying prompts. Experimental results on four datasets for SAM and its two variant models demonstrate the powerful attack capability and transferability of DarkSAM. Our codes are available at: https://***/CGCL-codes/DarkSAM. © 2024 Neural information processing systems foundation. All rights reserved.

关键词：

来源：评论

学校读者我要写书评

暂无评论

A Method for Facilitating Inventive Design Based on Semantic Similarity and Case-Based Reasoning

引用

Procedia Engineering 2015年 131卷 194-203页

作者： Wei Yan Cecilia Zanni-Merk François Rousselot Denis Cavallucci School of Information Science and Engineering Shandong Normal University Jinan 250014 China Shandong Provincial Key Laboratory for Novel Distributed Computer Software Technology Jinan 250014 China LGECO / INSA Strasbourg 24 Boulevard de la Victoire 67084 Strasbourg Cedex France LSIIT / FDBT Team (UMR CNRS 7005) – Pôle API BP 10413 67412 Illkirch Cedex France

While TRIZ is increasingly developing both in research and education, new users always encounter difficulties in their first attempts to practice it. In such situation, Altshuller's original matrix often appears as an “easy-to-begin- with” tool. However, while not being representative of what TRIZ really is, it continues to seduce new users, teachers and trainers. Several approaches to automate the use of the contradiction matrix have been proposed in literature, and researches on the automatic match between specific parameters of the artefact being considered and Altshuller's generalized parameters remains a topic of interest. In this paper, we propose a way to automate the proposal of equivalencies between a specific parameter and a generic parameter from the matrix. This method calculates the semantic distance between short texts, and uses it to fill the semantic gap between the specific parameters of the artefact and the generalized ones. Case-based reasoning is also used to improve the whole accuracy of the method.

关键词： Ontology Semantic distance Information content WordNet Case-based reasoning

来源：评论

学校读者我要写书评

暂无评论

A Heuristic Method of Using the Pointers to Physical Effects in Su-Field Analysis

引用

Procedia Engineering 2015年 131卷 539-550页

作者： Wei Yan Cecilia Zanni-Merk François Rousselot Denis Cavallucci Pierre Collet School of Information Science and Engineering Shandong Normal University Jinan 250014 China Shandong Provincial Key Laboratory for Novel Distributed Computer Software Technology Jinan 250014 China LGECO / INSA Strasbourg 24 Boulevard de la Victoire 67084 Strasbourg Cedex France LSIIT / BFO Team (UMR CNRS 7005) – Pôle API BP 10413 67412 Illkirch Cedex France

While TRIZ is increasingly developing both in research and education, users always encounter difficulties in their attempts to practice it, especially for one of the most complicated inventive problem solving tools, the Su-Field analysis, which is used to analyze and improve the efficiency of a technical system. Generally, the process of using a Su-Field model to solve a specific inventive problem includes: building a Problem Model, mapping it to a Generic Problem Model, finding a Generic Solution Model based on the corresponding inventive standard and finally establishing and interpreting a Solution Model in real life. As one of the most important phases of the Su-Field analysis, the last step is normally implemented manually with the help of the pointers to physical effects, which link generic technical functions to specific applications and systems. The adequate pointer, compatible with the context of the specific problem, should be chosen to assist the users to instantiate the Solution Model. However, the pointers to physical effects and the specific problems are built at different levels of abstraction, so it is difficult for the users to choose among many eligible pointers to physical effects given a certain function. This paper proposes a heuristic method to use the pointers to physical effects based on ontology reasoning. The use of formal models, called ontologies, allows to formally define the concepts about the models and the effects and to apply logical reasoning on them. An inventive standards ontology and a physical effects ontology are firstly built to describe the process of Su- Field analysis, and then the logical reasoning rules are established according to a lexical dictionary, the analysis of the inventive standards and the physical effects. The lexical dictionary, based on WordNet, is built to improve the selection of the appropriate effect to apply. Finally, ontology reasoning is implemented to provide heuristic physical effects for the users.

关键词： Su-Field Analysis Inventive Standards Pointers to Physical Effects WordNet Ontology Reasoning

来源：评论

学校读者我要写书评

暂无评论

Vehicle Destination Prediction Based on Trajectory Data

Vehicle Destination Prediction Based on Trajectory Data

引用

Neural Networks, Information and Communication Engineering (NNICE), International Conference on

作者： Wenming Guo Ruiqi Jia Che Zong Ruichao Li Key Laboratory of Trustworthy Distributed Computing and Service Ministry of Education School of Computer Science (National Pilot Software Engineering School) Beijing University of Posts and Telecommunications Beijing China School of Information Engineering Xinjiang Institute of Engineering Urumqi China School of Artificial Intelligence Beijing University of Posts and Telecommunications Beijing China

With the increasing requirements of real-time traffic management in urban transportation, vehicle destination prediction data, as an important part of smart city brain data, is becoming more and more important in urban traffic management and operation. This paper builds a vehicle destination prediction scheme based on trajectory data. By collecting public vehicle trajectory datasets, the Sequence to Sequence (Seq2Seq) sequence generation model and the Text-Convolutional Neural Network (Text-CNN) sequence classification model are built respectively to implement the vehicle destination prediction. In order to evaluate the performance of the models, Accuracy and Mean Distance are used as metrics to compare the model performance under different training data conditions. Advantage analysis was done in combination with different application scenarios for practical application reference. According to the prediction performance of different models built in this work in different application scenarios, the advantages are analyzed for reference in practice.

关键词：

来源：评论

学校读者我要写书评

暂无评论

X-Prcaf : Xposed Based Protecting Cache File from Leaks in Android Social Applications

X-Prcaf : Xposed Based Protecting Cache File from Leaks in A...

引用

International Conference on Trustworthy Systems and their Applications (TSA)

作者： Qi Tang Wen Zhang Xiaoyong Li Bin Wang Key Laboratory of Trustworthy Distributed Computing and Service (BUPT) Ministry of Education Beijing University of Posts and Telecommunications Beijing P.R. China Beijing Key Lab of Intelligent Telecommunication Software and Multimedia Beijing University of Posts and Telecommunications Beijing P.R. China School of Computer Science Beijing University of Posts and Telecommunications Beijing P.R. China

ISBN: (纸本)9781509035403

With the high frequent use of social applications on Android platform, the cache file privacy disclosure issues have become increasingly serious. To our best knowledge, there is no effective privacy protection solution for social applications cache files. In this paper, we analyze the present situation of social applications cache file leaks on Android platform, and provide a privacy disclosure assessment criterion based on file storage directories and security state machines. And a cache file privacy protection framework, X-Prcaf (Xposed-based-Protecting-Cache-File), is proposed, which can make social applications avoid privacy data leaks in running process. This framework mainly uses taint tracking technology, operating system hook technology, and cryptographic technology. It aims to protect the entire life cycle of the social applications cache files, by strategy pre-generation, real-time monitoring and security reinforcement. Experiments demonstrate that X-Prcaf has a good effect on the cache file leaks of social software.

关键词： Data privacy software Cryptography Smart phones Privacy Telecommunications

来源：评论

学校读者我要写书评

暂无评论

Penetration of authority under cloud environment based on RBAC

引用

Huazhong Keji Daxue Xuebao (Ziran Kexue Ban)/Journal of Huazhong University of Science and Technology (Natural Science Edition) 2013年第SUPPL.2期41卷 68-71页

作者： Zou, Jiashun Zhang, Yongsheng Gao, Yan School of Information Science and Engineering Shandong Normal University Jinan 250014 China Shandong Provincial Key Laboratory for Novel Distributed Computer Software Technology Shandong Normal University Jinan 250014 China

A research was made on the weak security of access control model based on the role, which was showed in the cross-domain access of cloud computing. An access control strategy based on the hops was put forward by introducing the role mapping technology and the definition of hop. It was presented to solve the problem of penetration of authority, which emerged in the process of role mapping. The strategy was studied from two aspects of both access process and formal definition, and according simulation experiment was done. Compared with other strategies, this strategy achieved fine-grained control on distance of cross-domain access by adding the corresponding field in the role set and user set. This strategy produced no new collection, and it did not need to establish any mirror role in the process of cross-domain access. The experimental results show that this strategy has higher efficiency and lower consumption of storage compared with the traditional strategy about cross-domain access.

关键词： Access control

来源：评论

学校读者我要写书评

暂无评论

Review of Semi-Structured Document Information Extraction Techniques Based on Deep Learning

Review of Semi-Structured Document Information Extraction Te...

引用

Machine Learning, Cloud Computing and Intelligent Mining (MLCCIM), International Conference on

作者： Yangchun Li Wei Jiang Shouyou Song Chinese Academy of Cybersapce Studies School of Computer Science (National Pilot Software Engineering School) Beijing University of Posts and Telecommunications Beijing China Chinese Academy of Cybersapce Studies Beijing China Key Laboratory of Trustworthy Distributed Computing and Service (BUPT) Chinese Academy of Cybersapce Studies Ministry of Education Beijing University of Posts and Telecommunications Beijing China

With the advent of global digital transformation, using an intelligent method based on deep learning to extract crucial information from semi-structured documents, as represented by various types of receipts and invoices, has emerged as an imperative measure to ensure business stability, data security, and improved work efficiency. This paper provides a detailed review on deep learning-based techniques for information extraction, with systematic introduction, hierarchical analysis, method comparison, and summary with expectations for future development. The review begins with a comprehensive explication of the defining characteristics of semi-structured documents, along with a detailed introduction to the research background, application areas, and technical challenges related to information extraction from semi-structured documents. Then the review extends to an overview of two developmental stages, i.e. the shift from traditional information extraction to deep learning-based information extraction, followed by discussion about technical architecture and method classification, which elaborates on key technologies in terms of typical datasets, detection and recognition, and information reduction. Lastly, paper summarizes the prospects and development in the field. Future research will focus on strengthening algorithm universal and lightweight, as well as improving information protection capabilities and the diversity of datasets.

关键词：

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：