检索结果-内蒙古大学图书馆

Proceedings of the Digital Image Computing: Technqiues and Applications (DICTA)

作者： Yiheng Lyu Lian Xu Mohammed Bennamoun Farid Boussaid Girish Dwivedi Department of Computer Science and Software Engineering The University of Western Australia Perth Australia Harry Perkins Institute of Medical Research Perth Australia Department of Electrical Electronic and Computer Engineering The University of Western Australia Perth Australia Medical School The University of Western Australia Perth Australia Department of Cardiology Fiona Stanley Hospital Perth Australia

ISBN: (数字)9798350379037

ISBN: (纸本)9798350379044

Weakly supervised brain tumor segmentation is a critical challenge in medical image analysis, characterized by inherent biological complexity and the scarcity of annotations. For segmentation on magnetic resonance imaging from image-level annotations, the intrinsic intra-class heterogeneity hampers accurate segmentation on distal images. In this paper, we propose a transformer-based model for weakly supervised brain tumor segmentation. Our approach tackles intra-class heterogeneity by utilizing an Importance-Aware Class Prediction strategy to improve segmentation accuracy on positive images. Additionally, we integrate Affine Transformation Equivariant Regularization to ensure consistent tumor localization. Finally, we introduce a Cross-Plane Refinement method to refine inaccurate segmentations using high-confidence images within the same volume. Evaluated on the Brain Tumor Segmentation 2023 dataset, our model demonstrates state-of-the-art performance on native Tl and T2-weighted sequences.

关键词： Location awareness Image segmentation Accuracy Annotations Magnetic resonance imaging Semantic segmentation Semantics Brain tumors Brain modeling Transformers

来源：评论

学校读者我要写书评

暂无评论

How re-sampling helps for long-tail learning? 23

How re-sampling helps for long-tail learning?

引用

Proceedings of the 37th International Conference on Neural Information Processing Systems

作者： Jiang-Xin Shi Tong Wei Yuke Xiang Yu-Feng Li National Key Laboratory for Novel Software Technology Nanjing University Nanjing China School of Computer Science and Engineering Southeast University Nanjing China Consumer BG Huawei Technologies Shenzhen China

Long-tail learning has received significant attention in recent years due to the challenge it poses with extremely imbalanced datasets. In these datasets, only a few classes (known as the head classes) have an adequate number of training samples, while the rest of the classes (known as the tail classes) are infrequent in the training data. Re-sampling is a classical and widely used approach for addressing class imbalance issues. Unfortunately, recent studies claim that re-sampling brings negligible performance improvements in modern long-tail learning tasks. This paper aims to investigate this phenomenon systematically. Our research shows that re-sampling can considerably improve generalization when the training images do not contain semantically irrelevant contexts. In other scenarios, however, it can learn unexpected spurious correlations between irrelevant contexts and target labels. We design experiments on two homogeneous datasets, one containing irrelevant context and the other not, to confirm our findings. To prevent the learning of spurious correlations, we propose a new context shift augmentation module that generates diverse training images for the tail class by maintaining a context bank extracted from the head-class images. Experiments demonstrate that our proposed module can boost the generalization and outperform other approaches, including class-balanced re-sampling, decoupled classifier re-training, and data augmentation methods. The source code is available at https://***/code_***.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Understanding and Detecting Privacy Leakage Vulnerabilities in Hyperledger Fabric Chaincodes

Understanding and Detecting Privacy Leakage Vulnerabilities ...

引用

International Symposium on software Reliability engineering (ISSRE)

作者： Ziming Chen Yue Li Jianbo Gao Jiashuo Zhang Ke Wang Jianbin Hu Zhi Guan Zhong Chen School of Computer Science Peking University Beijing China College of Computer Science and Technology Taiyuan University of Technology Taiyuan China Beijing Key Laboratory of Security and Privacy in Intelligent Transportation Beijing Jiaotong University Beijing China Peking University Chongqing Research Institute of Big Data Chongqing China National Engineering Research Center for Software Engineering Peking University Beijing China

ISBN: (数字)9798350353884

ISBN: (纸本)9798350353891

The application on a blockchain cannot maintain secrecy because its data is replicated across all peers in the network. To remedy this problem, Hyperledger Fabric introduces private data collection (PDC) into its smart contract (i.e. chaincode) to facilitate applications that require privacy. However, recent studies have revealed that PDC is too complex for chaincode developers to fully understand and use correctly, leading to privacy leaks vulnerabilities. In this paper, we present an empirical study on the prevalence of PDC misuse in chaincodes by extracting privacy leakage cases from StackOverflow posts and Hyperledger Fabric repositories on GitHub. Subsequently, we systematically categorize the misuse of PDC into three categories of vulnerabilities resulting in the leakage of private data and provide formal definitions for them. Furthermore, we develop PDChecker, an automated security analysis framework for identifying the privacy and security vulnerabilities in Fabric chaincodes. We evaluated PDChecker on 956 real-world chaincodes applying PDC and found that 67.78% of them contain at least one privacy leakage vulnerability. In addition, PDChecker uncovered 10 zero-day vulnerabilities documented by the China National Vulnerability Database.

关键词： Privacy Data privacy Distributed ledger Databases Smart contracts Fabrics Blockchains software reliability Security software development management

来源：评论

学校读者我要写书评

暂无评论

Scribble-supervised semantic segmentation with prototype-based feature augmentation 24

Scribble-supervised semantic segmentation with prototype-bas...

引用

Proceedings of the 41st International Conference on Machine Learning

作者： Guiyang Chan Pengcheng Zhang Hai Dong Shunhui Ji Bainian Chen Key Laboratory of Water Big Data Technology of Ministry of Water Resources Hohai University Nanjing China and College of Computer Science and Software Engineering Hohai University Nanjing China School of Computing Technologies RMIT University Melbourne Australia

Scribble-supervised semantic segmentation presents a cost-effective training method that utilizes annotations generated through scribbling. It is valued in attaining high performance while minimizing annotation costs, which has made it highly regarded among researchers. Scribble supervision propagates information from labeled pixels to the surrounding unlabeled pixels, enabling semantic segmentation for the entire image. However, existing methods often ignore the features of classified pixels during feature propagation. To address these limitations, this paper proposes a prototype-based feature augmentation method that leverages feature prototypes to augment scribble supervision. Experimental results demonstrate that our approach achieves state-of-the-art performance on the PASCAL VOC 2012 dataset in scribble-supervised semantic segmentation tasks. The code is available at https://***/TranquilChan/PFA.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Capability-based IoT access control using blockchain

引用

Digital Communications and Networks 2021年第4期7卷 463-469页

作者： Yue Liu Qinghua Lu Shiping Chen Qiang Qu Hugo’Connor Kim-Kwang Raymond Choo He Zhang School of Computer Science and Engineering University of New South WalesAustralia Data61 CSIROAustralia Shenzhen Institutes of Advanced Technology Chinese Academy of SciencesChina Department of Information Systems and Cyber Security University of Texas at San AntonioUSA Software Institute Nanjing UniversityChina

Internet of Things(IoT)devices facilitate intelligent service delivery in a broad range of settings,such as smart offices,homes and ***,the existing IoT access control solutions are mainly based on conventional identity management schemes and use centralized *** are knowm security and privacy limitations with such schemes and architectures,such as the single-point failure or surveillance(e.g.,device tracking).Hence,in this paper,we present an architecture for capability-based IoT access control utilizing the blockchain and decentralized identifiers to manage the identity and access control for IoT ***,we propose a protocol to provide a systematic view of system interactions,to improve *** also implement a proof-of-concept prototype of the proposed approach and evaluate the prototype using a real-world use *** evaluation results show that the proposed solution is feasible,secure,and scalable.

关键词： Blockchain Internet of things Capability-based access control Identity management Architecture design

来源：评论

学校读者我要写书评

暂无评论

FGRL-Net: Fine-Grained Personalized Patient Representation Learning for Clinical Risk Prediction Based on EHRs

FGRL-Net: Fine-Grained Personalized Patient Representation L...

引用

IEEE International Conference on Systems, Man and Cybernetics

作者： KaKit Chio Wenhao Zhu Lihua He Dian Zhang Xu Yang Wuman Luo Faculty of Applied Sciences Macao Polytechnic University Macao S.A.R China School of Information and Communication Engineering University of Electronic Science and Technology of China Sichuan Province China Department of Computer Science and Software Engineering Shenzhen University Shenzhen China

Personalized patient representation learning (PPRL) is a critical element in clinical risk prediction. It aims to obtain a complete portrait of each patient based on Electronic Health Records (EHR). Although existing works have achieved remarkable progress in healthcare prediction, there are still three major issues. First, feature correlation is crucial for risk prediction, but it has not yet been fully exploited by existing works. Second, variation pattern of dynamic feature contains useful information about patient's physical status, but adaptive pattern recognition is still a challenge. Third, existing works usually adopt a two-stage embedding process to process each dimension of the EHR data. However, some useful low-level information for PPRL will be lost. To address these issues, in this paper, we propose a fine-grained PPRL architecture named FG RL- N et for clinical risk prediction based on EHR. Specifically, we propose a Medical Feature Correlation Detection Module (FCM) to effectively learn the feature correlations for each patient and a Temporal Variation Pattern Recognition Module (TVM) to effectively detect the variation patterns of each dynamic feature. Moreover, we design a Fine-Grained Representation Mechanism (FGRM) to preserve the low-level information (from both feature and visit dimensions) useful for risk prediction. In addition, in the stage of data preprocessing, We utilize generic medical classification knowledge to classify numerical dynamic data. We conduct the in-hospital mortality experiment and the decompensation experiment on a real-world dataset. The experiment results show that the FGRL-Net outperforms state-of-the-art approaches. The source code is provided in github https://***/JackyChio/FGRL-Net.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Research on Malware software Detection Technology Based on SVM on Cloud Platform for Mobile Terminal

Research on Malware Software Detection Technology Based on S...

引用

International Conference on Artificial Intelligence and computer Information Technology (AICIT)

作者： Liming Lin Ye Yao JiaYu Li Jia Yao Wending Chen Siyang Huang School of Software Northwestern Polytechnical University Xi'an City China School of Computer Science Northwestern Polytechnical University Xi'an City China Queen Mary University of London Engineering School NPU Northwestern Polytechnical University Xi'an City China

In recent years, the number of new malware samples has shown a trend of rapid growth. The increasing number of malware has resulted in a huge number of signature databases that the host needs to save, which is difficult to maintain. In view of the above situation, this article builds a distributed cloud platform based on Hadoop, and uses the related technology of the cloud platform to deal with the rapid growth of the number of malware. At the same time, the use of the distributed cloud platform greatly improves the detection efficiency. In order to solve the problem of low detection accuracy of unknown software, a malware detection method based on API call sequence is proposed. First use AndroGuard to extract all the function call graphs under the package of the main class in the APK file. Then, through the analysis of the function call graph, the API call sequence of all the entry nodes is obtained; further, the sequence is sorted according to the address information of the first node of each sequence. Splice the sorted sequences in order, and the resulting sequence is the time-series-based API call sequence. After dimensionality reduction through AutoEncoder, machine learning algorithm SVM is used for malware detection. Experimental results show that the accuracy of the SVM detection model based on the API call sequence is higher than that of the SVM detection model based on other features and other machine learning detection models based on the API call sequence, and can detect Android malware more effectively.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Boosting Adversarial Transferability across Model Genus by Deformation-Constrained Warping

arXiv

引用

arXiv 2024年

作者： Lin, Qinliang Luo, Cheng Niu, Zenghao He, Xilin Xie, Weicheng Hou, Yuanbo Shen, Linlin Song, Siyang Computer Vision Institute School of Computer Science & Software Engineering Shenzhen University China Shenzhen Institute of Artificial Intelligence and Robotics for Society China Guangdong Key Laboratory of Intelligent Information Processing China WAVES Research Group Ghent University Belgium University of Leicester United Kingdom

Adversarial examples generated by a surrogate model typically exhibit limited transferability to unknown target systems. To address this problem, many transferability enhancement approaches (e.g., input transformation and model augmentation) have been proposed. However, they show poor performances in attacking systems having different model genera from the surrogate model. In this paper, we propose a novel and generic attacking strategy, called Deformation-Constrained Warping Attack (DeCoWA), that can be effectively applied to cross model genus attack. Specifically, DeCoWA firstly augments input examples via an elastic deformation, namely Deformation-Constrained Warping (DeCoW), to obtain rich local details of the augmented input. To avoid severe distortion of global semantics led by random deformation, DeCoW further constrains the strength and direction of the warping transformation by a novel adaptive control strategy. Extensive experiments demonstrate that the transferable examples crafted by our DeCoWA on CNN surrogates can significantly hinder the performance of Transformers (and vice versa) on various tasks, including image classification, video action recognition, and audio recognition. Code is made available at https://***/LinQinLiang/DeCoWA. © 2024, CC BY-NC-SA.

关键词： Semantics

来源：评论

学校读者我要写书评

暂无评论

PRAD: Unsupervised KPI Anomaly Detection by Joint Prediction and Reconstruction of Multivariate Time Series 19

PRAD: Unsupervised KPI Anomaly Detection by Joint Prediction...

引用

2022 IEEE SmartWorld, 19th IEEE International Conference on Ubiquitous Intelligence and Computing, 2022 IEEE International Conference on Autonomous and Trusted Vehicles Conference, 22nd IEEE International Conference on Scalable Computing and Communications, 2022 IEEE International Conference on Digital Twin, 8th IEEE International Conference on Privacy Computing and 2022 IEEE International Conference on Metaverse, SmartWorld/UIC/ATC/ScalCom/DigitalTwin/PriComp/Metaverse 2022

作者： Xiong, Zhiying Fan, Qilin Wang, Kai Li, Xiuhua Zhang, Xu Xiong, Qingyu Chongqing University School of Big Data and Software Engineering Chongqing China Chongqing University Chongqing Key Laboratory of Digital Cinema Art Theory and Technology Chongqing China Harbin Institute of Technology School of Computer Science and Technology Weihai China Harbin Institute of Technology Research Institute of Cyberspace Security Weihai China Haihe Laboratory of Information Technology Application Innovation Tianjin China University of Exeter Department of Computer Science Exeter United Kingdom

ISBN: (纸本)9798350346558

Detecting anomalies for key performance indicator (KPI) data is of paramount importance to ensure the quality and reliability of network services. However, building the anomaly detection system for KPI is challenging due to the scarcity of abnormal labels and highly dynamic or even unseen patterns. In this paper, we propose an unsupervised KPI anomaly detection framework named PRAD by jointly optimizing prediction-based and reconstruction-based modules. Specifically, PRAD employs two parallel graph attention networks (GATs) to learn metric-oriented and time-oriented relationships. A temporal convolutional attention network (TCAN) is used to capture complex long-term dependencies. Furthermore, to tackle the overfitting issue in the variational auto-encoder (VAE), PRAD designs a reconstruction-based module that combines VAE with a generative adversarial network (GAN). The experimental results on the two public datasets demonstrate that the proposed PRAD outperforms existing anomaly detection methods. © 2022 IEEE.

关键词： Anomaly detection

来源：评论

学校读者我要写书评

暂无评论

Analysis of vocational education dynamics evolution based on topic modeling

Analysis of vocational education dynamics evolution based on...

引用

IEEE International Symposium on Information (IT) in Medicine and Education, ITME

作者： Xinhua Wang Xiaomei Yu Yunmeng Jiang Xiangwei Zheng Wei Li School of Information Science and Engineering Shandong Normal University Jinan China Shandong Provincial Key Laboratory for Distributed Computer Software Novel Technology Jinan China Shandong Normal University Library Shandong Normal University Jinan China

ISBN: (数字)9798350356670

ISBN: (纸本)9798350356687

Clarifying the focus of vocational education dynamics is crucial for understanding the current situation and future development trends. However, comprehensive exploration on vocational education dynamics is inadequate presently. In this paper, textual information on vocational education is analyzed with Latent Dirichlet Allocation, BERTopic, and the Dynamic Topic Model. Two key issues are thoroughly investigated as follows: (1) What are the main aspects of the current development of vocational education; (2) How to learn the embeddings of vocational education text and further to promote the formulation of policy. The experiments are conducted and the experimental results show that LDA and BERTopic model achieve effective performance in clustering the texts of vocational education dynamics, and the DTM captures the topics trend of vocational education, which facilitate the government to guide the formulation and implementation of policies.

关键词： Analytical models Education Government Market research Dynamic scheduling Resource management Information technology

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：