检索结果-内蒙古大学图书馆

8th International Conference on Computer Science and Artificial Intelligence, CSAI 2024

作者： Duan, Lijuan Zhang, Zichen Gong, Zhi Beijing Key Laboratory of Trusted Computing Beijing University of Technology Beijing China School of Computer Science Beijing University of Technology Beijing China National Engineering Laboratory for Key Technologies of Information Security Level Protection Beijing China

ISBN: (纸本)9798400718182

Recent deep learning models have advanced object detection capabilities, but their reliance on labeled data limits scalability. Fine-grained zero-shot detection remains challenging due to the need to generalize to unseen classes and distinguish subtle features. To address this, we propose Fine-Grained Zero-Shot Object Detection via Semantic Decoupling (ZSD-SD). Our approach introduces a semantic decoupling module to separate visual features into semantic-related and semantic-unrelated components, facilitating alignment between visual and semantic spaces. A semantic-visual learning module further decouples these features, enhancing visual representations relevant to external semantic knowledge. This semantic-related representation significantly improves zero-shot performance. Experiments on two fine-grained object detection datasets demonstrate that ZSD-SD achieves state-of-the-art results. © 2024 Copyright held by the owner/author(s).

关键词： Zero-shot learning

来源：评论

学校读者我要写书评

暂无评论

Burst image super-resolution based on dual branch fusion and adaptive frame selection

引用

Displays 2025年 89卷

作者： Duan, Lijuan Zuo, Wendi Gu, Ke Gong, Zhi School of Computer Science Beijing University of Technology Beijing100124 China Beijing Key Laboratory of Trusted Computing Beijing University of Technology Beijing100124 China National Engineering Laboratory for Critical Technologies of Information Security Classified Protection Beijing University of Technology Beijing100124 China School of Information Science and Technology Beijing University of Technology Beijing100124 China Chongqing Research Institute of Beijing University of Technology China

The modern handheld camera is capable of rapidly capturing multiple images and subsequently merging them into a single image. The extant methodologies typically select the first frame as the reference frame, utilising the information from the remaining frames and the information from the reference frame to calculate the high-resolution image. However, for complex scenes and unstable shooting situations, this fixed frame selection is not the optimum solution. In this direction, we propose an adaptive frame selection method which, by calculating the frame channel weight information, selects the best frame image as the reference frame to perform the subsequent computation. Moreover, to enhance the visual quality of high-resolution images, we propose a dual-branch fusion module in the feature fusion phase for the sequence attributes of the input frame images. This allows the network to concentrate on the temporal global features of the input sequence and the spatial local detail features of the frame images. Subsequently, the feature map is obtained through residual computation utilising the adaptive selected frame image and the image features obtained through fusion. The image is then reconstructed through up-sampling to obtain a high-resolution image. The experimental results on the BurstSR and RealBSR datasets demonstrate that our approach not only outperforms existing techniques in terms of evaluation metrics but also exhibits superior visual effects. © 2025

关键词： Photointerpretation

来源：评论

学校读者我要写书评

暂无评论

TPTVer： A Trusted Third Party Based Trusted Verifier for Multi-Layered Outsourced Big Data System in Cloud Environment

引用

China Communications 2018年第2期15卷 122-137页

作者： Jing Zhan Xudong Fan Lei Cai Yaqi Gao Junxi Zhuang College of Computer Science Faculty of Information Technology Beijing University of Technology Beijing China Beijing Key Laboratory of Trusted Computing Beijing China National Engineering Laboratory for Critical Technologies of Information Security Classified Protection Beijing China

Cloud computing is very useful for big data owner who doesn't want to manage IT infrastructure and big data technique details. However, it is hard for big data owner to trust multi-layer outsourced big data system in cloud environment and to verify which outsourced service leads to the problem. Similarly, the cloud service provider cannot simply trust the data computation applications. At last,the verification data itself may also leak the sensitive information from the cloud service provider and data owner. We propose a new three-level definition of the verification, threat model, corresponding trusted policies based on different roles for outsourced big data system in cloud. We also provide two policy enforcement methods for building trusted data computation environment by measuring both the Map Reduce application and its behaviors based on trusted computing and aspect-oriented programming. To prevent sensitive information leakage from verification process,we provide a privacy-preserved verification method. Finally, we implement the TPTVer, a Trusted third Party based Trusted Verifier as a proof of concept system. Our evaluation and analysis show that TPTVer can provide trusted verification for multi-layered outsourced big data system in the cloud with low overhead.

关键词： big data security outsourced ser-vice security MapReduce behavior trustedverification trusted third party

来源：评论

学校读者我要写书评

暂无评论

A trusted behaviour measurement scheme based on feedback and control in trusted network

引用

International Journal of Wireless and Mobile Computing 2021年第2期20卷 99-106页

作者： Zhuang, Junxi Zhao, Xu Gong, Bei Zhan, Jing College of Computer Science Beijing Key Laboratory of Trusted Computing Beijing University of Technology Beijing100124 China National Engineering Laboratory for Critical Technologies of Information Security Classified Protection Beijing University of Technology Beijing100124 China

To solve the real-time trusted measurement of user behaviour in trusted networks, a new combination of dynamic and static, subjective and objective behaviour trusted measurement scheme is presented. The scheme analyses the credibility of the real-time user state and the direct trusted measurement value of the user behaviour, and finally controls the user behaviour by the feedback measurement of the user behaviour. The research results show that when the ratio of untrusted nodes is 15%, 30% and 50%, the trusted detection rate of this paper can reach more than 92%, and the untrusted node in trusted network can be identified with the increase of time. Compared with traditional measurement scheme, the scheme can be more accurate and reasonable to calculate the credibility of user behaviour, and can reflect the changes of the credibility of users’ actual behaviour at all times, and ultimately implement the feedback control on user behaviour. Copyright © 2021 Inderscience Enterprises Ltd.

关键词： Feedback control

来源：评论

学校读者我要写书评

暂无评论

A novel image encryption method based on hyperchaotic system and pixel information association

引用

International Journal of Wireless and Mobile Computing 2018年第3期15卷 207-214页

作者： Duan, Lijuan Zhang, Dongkui Ning, Zhenhu Xu, Fan Cui, Guoqin Liang, Peng Faculty of Information Technology Beijing University of Technology Beijing Key Laboratory of Trusted Computing National Engineering Laboratory for Critical Technologies of Information Security Classified Protection Beijing100124 China State Key Laboratory of Digital Multi-Media Chip Technology Vimicro Corporation Beijing100191 China State Information Center National Engineering Laboratory for e-Government Integration and Application Beijing100045 China

In recent years, a variety of image encryption methods based on chaotic systems has been proposed, and has achieved good results. In order to improve the security of image encryption, this paper proposes a new image encryption method with a dynamic key, based on a hyperchaotic system and the association of pixel information, and designs the three-level encryption structure with diffusion encryption, scrambling encryption and re-diffusion encryption. The innovation of this method is to add the whole information of the image and the position information of the pixels in the diffusion encryption, and add the data information of the pixels in the scrambling encryption. The plaintext information of the image is added to the encryption process, which can enhance the security of encryption. Experiments demonstrate that the image encryption method proposed in this paper has the characteristics of strong sensitivity, low correlation between adjacent pixels, anti-differential attack and high security. Copyright © 2018 Inderscience Enterprises Ltd.

关键词： Cryptography

来源：评论

学校读者我要写书评

暂无评论

Knowledge Graph Embedding Method Based on Entity Metric Learning 23

Knowledge Graph Embedding Method Based on Entity Metric Lear...

引用

9th International Conference on Computing and Artificial Intelligence, ICCAI 2023

作者： He, Meng Duan, Lijuan Zhang, Baochang Han, Shengwen Faculty of Information Technology Beijing University of Technology Beijing Key Laboratory of Trusted Computing National Engineering Laboratory for Key Technologies of Information Security Level Protection China Institute of Artificial Intelligence Beihang University China Faculty of Information Technology Beijing University of Technology China

ISBN: (纸本)9781450399029

A knowledge graph is a knowledge base of graph structure that stores massive facts and their relations in the form of triplets. Embedding entities and relations in knowledge graph into low-dimensional continuous vector space is of great significance for downstream tasks, whereas most existing methods give an end-to-end fixed representation. Based on deep metric learning, we give different embeddings for the same entity in different relations and positions, thus dynamically representing entities in knowledge graph and effectively modeling complex relations. In addition, a path-based negative sampling method is proposed to widen the distance between similar entities to better deal with the aggregation problem of embedding entities in the representation space under one-to-many relation. Experiments show that the MRR and Hit@1 of this method on the linked prediction dataset FB15k-237 reach 0.360 and 0.267 respectively, which are higher than the mainstream distance models and semantic matching models. Finally, the MRR-Dimension curves demonstrates that our model can perform well in both low and high embedding dimensions and has the potential to be applied to large-scale knowledge graph. © 2023 Copyright held by the owner/author(s).

关键词： Knowledge graph

来源：评论

学校读者我要写书评

暂无评论

A static technique for detecting input validation vulnerabilities in Android apps

引用

Science China(information Sciences) 2017年第5期60卷 210-225页

作者： Zhejun FANG Qixu LIU Yuqing ZHANG Kai WANG Zhiqiang WANG Qianru WU National Computer Network Intrusion Protection Center University of Chinese Academy of Sciences National Computer Network Emergency Response Technical Team/Coordination Center of China State Key Laboratory of Information Security Institute of Information EngineeringChinese Academy of Sciences Beijing Electronic Science and Technology Institute

Input validation vulnerabilities are common in Android apps, especially in inter-component communications. Malicious attacks can exploit this kind of vulnerability to bypass Android security mechanism and compromise the integrity, confidentiality and availability of Android devices. However, so far there is not a sound approach at the source code level for app developers aiming to detect input validation vulnerabilities in Android apps. In this paper, we propose a novel approach for detecting input validation flaws in Android apps and we implement a prototype named Easy IVD, which provides practical static analysis of Java source *** IVD leverages backward program slicing to extract transaction and constraint slices from Java source *** Easy IVD validates these slices with predefined security rules to detect vulnerabilities in a known *** detect vulnerabilities in an unknown pattern, Easy IVD extracts implicit security specifications as frequent patterns from the duplicated slices and verifies them. Then Easy IVD semi-automatically confirms the suspicious rule violations and reports the confirmed ones as vulnerabilities. We evaluate Easy IVD on four versions of original Android apps spanning from version 2.2 to 5.0. It detects 58 vulnerabilities including confused deputy attacks and denial of service attacks. Our results prove that Easy IVD can provide a practical defensive solution for app developers.

关键词： input validation static analysis program slicing vulnerability detection Android security

来源：评论

学校读者我要写书评

暂无评论

Learning EEG synchronization patterns for epileptic seizure prediction using bag-of-wave features

引用

Journal of Ambient Intelligence and Humanized Computing 2023年第11期14卷 15557-15572页

作者： Cui, Song Duan, Lijuan Qiao, Yuanhua Xiao, Ying Faculty of Information Technology Beijing University of Technology Beijing100124 China Beijing Key Laboratory of Trusted Computing National Engineering Laboratory for Critical Technologies of Information Security Classified Protection Beijing100124 China College of Applied Sciences Beijing University of Technology Beijing100124 China

Epileptic seizure prediction has the potential to promote epilepsy care and treatment. However, the seizure prediction accuracy does not satisfy the application requirements. In this paper, a novel framework for seizure prediction is proposed by learning synchronization patterns. For better representation, bag-of-wave (BoWav) feature extraction is proposed for modeling synchronization pattern of electroencephalogram (EEG) signal. An interictal codebook and preictal codebook, representing the local segments, are constructed by a clustering algorithm. Within a period of EEG signal on all electrodes, local segments are projected onto the learned codebooks. The proposed feature expresses the synchronization pattern of EEG signal with the histogram feature. Moreover, extreme learning machine (ELM) is used to classify the sequence of features. Experiments are performed on the Kaggle seizure prediction challenge dataset and the CHB-MIT dataset. The experiment on the CHB-MIT achieves a sensitivity of 88.24% and a false prediction rate per hour of 0.25. © 2018, Springer-Verlag GmbH Germany, part of Springer Nature.

关键词： Feature extraction

来源：评论

学校读者我要写书评

暂无评论

EEG Source Localization Using Spatio-Temporal Neural Network

引用

China Communications 2019年第7期16卷 131-143页

作者： Song Cui Lijuan Duan Bei Gong Yuanhua Qiao Fan Xu Juncheng Chen Changming Wang Faculty of Information Technology Beijing University of Technology Beijing 100124 China Beijing Key Laboratory of Trusted Computing National Engineering Laboratory for Critical Technologies of Information Security Classified Protection Beijing 100124 China College of Applied Sciences Beijing University of Technology Beijing 100124 China The National Clinical Research Center for Mental Disorders & Beijing Key Laboratory of Mental Disorders Beijing Anding Hospital Capital Medical University Beijing 100088 China Advanced Innovation Center for Human Brain Protection Capital Medical University Beijing 100069 China

Source localization of focal electrical activity from scalp electroencephalogram (sEEG) signal is generally modeled as an inverse problem that is highly ill-posed. In this paper, a novel source localization method is proposed to model the EEG inverse problem using spatio-temporal long-short term memory recurrent neural networks (LSTM). The network model consists of two parts, sEEG encoding and source decoding, to model the sEEG signal and receive the regression of source location. As there does not exist enough annotated sEEG signals correspond to specific source locations, simulated data is generated with forward model using finite element method (FEM) to act as a part of training signals. A framework for source localization is proposed to estimate the source position based on simulated training data. Experiments are done on simulated testing data. The results on simulated data exhibit good robustness on noise signal, and the proposed network solves the EEG inverse problem with spatio-temporal deep network. The result show that the proposed method overcomes the highly ill-posed linear inverse problem with data driven learning.

关键词： electroencephalogram LSTM source localization spatio-temporal modeling

来源：评论

学校读者我要写书评

暂无评论

A federated sleep staging method based on adaptive re-aggregation and double prototype-contrastive using single-channel electroencephalogram

引用

engineering Applications of Artificial Intelligence 2025年 156卷

作者： Bian Ma Lijuan Duan Zhaoyang Huang Yuanhua Qiao College of Computer Science Beijing University of Technology Beijing 100124 China Beijing Key Laboratory of Trusted Computing Beijing 100124 China China National Engineering Laboratory for Critical Technologies of Information Security Classified Protection Beijing 100124 China Department of Neurology Xuanwu Hospital Capital Medical University Beijing 100053 China Beijing Key Laboratory of Neuromodulation Beijing 100053 China School of Mathematics Statistics and Mechanics Beijing University of Technology Beijing 100124 China

The limited availability of sleep data collected by individual institutions poses a significant challenge in developing automatic sleep staging models, especially given that most existing models rely heavily on data-driven approaches. The advent of federated learning has introduced an innovative and reliable paradigm for inter-institutional collaboration, facilitating task knowledge sharing and implicit data augmentation while preserving privacy. However, in traditional federated learning methodologies, local models are typically initialized with the global model at the onset of each iteration. This global-knowledge-centric approach often results in catastrophic forgetting during local training and inadequately adapts to the diversity of local data, compromising the overall generalization performance of the model. In this paper, we introduce a novel re-aggregation strategy for local model initialization that synergizes the global model with historical local models, thereby mitigating the undue influence of global knowledge and preserving local task-specific information. The re-aggregation weights are adaptively determined based on the confidence of the global and historical models on local datasets. Furthermore, to address the inherent ambiguity among sleep stages, a dual prototype-contrastive learning module is proposed, comprising Prototype-Consistency Contrastive (PCC) and Prototype-Distinguishability Contrastive (PDC) components. Specifically, the PCC component is designed to ensure consistency between local prototypes and unbiased prototypes derived from re-aggregation, enhancing intra-class knowledge coherence. The goal of the PDC component is to strengthen the discriminability among various sleep stages, improving inter-class differentiation. Comprehensive experiments conducted on two public datasets demonstrate the effectiveness, superiority, and flexibility of the proposed method.

关键词： Federated 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：