检索结果-内蒙古大学图书馆

Proceedings of the 38th International Conference on Neural Information Processing Systems

作者： Xiaoyuan Zhang Genghui Li Xi Lin Yichi Zhang Yifan Chen Qingfu Zhang Department of Computer Science City University of Hong Kong College of Computer Science and Software Engineering Shenzhen University Department of Statistics Indiana University Bloomingtom Departments of Mathematics and Computer Science Hong Kong Baptist University

ISBN: (纸本)9798331314385

Multiobjective optimization (MOO) plays a critical role in various real-world domains. A major challenge therein is generating K uniform Pareto-optimal solutions to approximate the entire Pareto front. To address this issue, this paper firstly introduces fill distance to evaluate the K design points, which provides a quantitative metric for the representativeness of the design. However, directly specifying the optimal design that minimizes the fill distance is nearly intractable due to the involved nested min - max - min problem structure. To address this, we propose a surrogate "max-packing" design for the fill distance design, which is easier to optimize and leads to a rate-optimal design with a fill distance at most 4 × the minimum value. Extensive experiments on synthetic and real-world benchmarks demonstrate that our proposed paradigm efficiently produces high-quality, representative solutions and outperforms baseline MOO methods.

关键词：

来源：评论

学校读者我要写书评

暂无评论

SKDLog: Self-Knowledge Distillation-based CNN for Abnormal Log Detection 19

SKDLog: Self-Knowledge Distillation-based CNN for Abnormal L...

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： Han, Ningning Lu, Siyang Wang, Dongdong Wang, Mingquan Tan, Xiaoman Wei, Xiang Beijing Jiaotong University School of Computer and Information Technology Beijing China University of Central Florida Department of Computer Science OrlandoFL United States Beijing Jiaotong University School of Software Engineering Beijing China Southwest University College of Computer and Information Science College of Software Chongqing China

ISBN: (纸本)9798350346558

System logs are critical to system health diagnosis, especially for big data distributed systems. To detect anomalies in logs, log analysis has become a popular and practical approach. However, with the increasing complexity of system logs in real-world systems, more noisy data are generated. The noise disturbs the regular detection procedure and decreases the detection quality. Hence, it is more difficult to improve detection performance under such uncertain circumstance. To tackle this challenge, we propose SKDLog, a novel and effective log anomaly detection method equipped with self-knowledge distillation. This approach leverages the knowledge from soft labels and refined feature maps through knowledge distillation. Moreover, it helps the detection model better acquire latent information. To better exploit feature maps, an auxiliary self-teacher branch is incorporated into the framework. After the integration, the model achieves performance gain in log anomaly detection. To demonstrate the effectiveness, we compare SKDLog with the state-of-the-art log-based anomaly detection approaches on HDFS and Hadoop Application datasets. Our model outperforms other approaches on both recall and F1-score. Furthermore, we challenge problem solving by performing experiments on an industry dataset and an unstable dataset. The experimental results show that SKDLog more accurately detects abnormal logs from noisy data with high recall and F1-score. © 2022 IEEE.

关键词： Distillation

来源：评论

学校读者我要写书评

暂无评论

Protein Residue Contact Prediction Based on Deep Learning and Massive Statistical Features from Multi-Sequence Alignment

引用

Tsinghua science and Technology 2022年第5期27卷 843-854页

作者： Huiling Zhang Min Hao Hao Wu Hing-Fung Ting Yihong Tang Wenhui Xi Yanjie Wei Shenzhen Institutes of Advanced Technology Chinese Academy of SciencesShenzhen 518055China University of Chinese Academy of Sciences Beijing 100049China College of Electronic and Information Engineering Southwest UniversityChongqing 400715China School of Software Engineering University of Science and Technology of ChinaHefei 230051China Department of Computer Science The University of Hong KongHong Kong 999077China School of Computer Science Beijing University of Posts and TelecommunicationsBeijing 100876China

Sequence-based protein tertiary structure prediction is of fundamental importance because the function of a protein ultimately depends on its 3 D *** accurate residue-residue contact map is one of the essential elements for current ab initio prediction protocols of 3 D structure ***,with the combination of deep learning and direct coupling techniques,the performance of residue contact prediction has achieved significant ***,a considerable number of current Deep-Learning(DL)-based prediction methods are usually time-consuming,mainly because they rely on different categories of data types and third-party *** this research,we transformed the complex biological problem into a pure computational problem through statistics and artificial *** have accordingly proposed a feature extraction method to obtain various categories of statistical information from only the multi-sequence alignment,followed by training a DL model for residue-residue contact prediction based on the massive statistical *** proposed method is robust in terms of different test sets,showed high reliability on model confidence score,could obtain high computational efficiency and achieve comparable prediction precisions with DL methods that relying on multi-source inputs.

关键词： multi-sequence alignment residue-residue contact prediction feature extraction statistical information Deep Learning(DL) high computational efficiency

来源：评论

学校读者我要写书评

暂无评论

Mutual Graph Learning Network and Diffusion Probabilistic Model-based Medical Image Segmentation

Mutual Graph Learning Network and Diffusion Probabilistic Mo...

引用

2023 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2023

作者： Su, Huaqiang Lei, Haijun Chen, Guoliang Chen, Xin Lei, Baiying Shenzhen University College of Computer Science & Software Engineering Shenzhen China Guangzhou First People's Hospital Department of Radiology Guangzhou China School of Biomedical Engineering Shenzhen University Medical School Shenzhen University Shenzhen China

ISBN: (纸本)9798350337488

Diffusion probabilistic models (DPM) can generate semantically valuable pixel-level representations and are widely used in medical image segmentation tasks. However, DPM faces challenges when dealing with medical image segmentation problems due to the irregular structure of medical images and the similarity between lesions and their surrounding environments. Therefore, this paper proposes a dual-branch Diff-UNet architecture to solve the medical image segmentation problem. Specifically, this architecture introduces the Transformer internal network on top of the standard UNet architecture based on DPM and realizes the interaction of UNet and Transformer branch features through bidirectional connection units to capture local features and remote dependencies better. In addition, through the feature fusion module (FFM), the global context information extracted by DPM is combined with the local detail features captured by the segmentation network. Simultaneously, this paper introduces a mutual graph learning (MGL) network to decompose the image into two task-specific feature maps, which are used to roughly locate the object position and capture the fine details of the object boundary. Finally, the cross attention (CA) module combines the edge information of the diffusion model with the features of the segmentation network to enhance the network's ability to perceive images. Experiments demonstrate the effectiveness of our Diff-UNet on challenging datasets, including self-collected databases and LUNA16. © 2023 IEEE.

关键词： Diffusion probability model feature fusion module medical segmentation mutual graph learning network

来源：评论

学校读者我要写书评

暂无评论

Adverse Drug Reaction Prediction and Feature Importance Mining Based on SIDER Dataset 3

Adverse Drug Reaction Prediction and Feature Importance Mini...

引用

3rd International Conference on Machine Learning and computer Application, ICMLCA 2022

作者： Chen, Tianqi Liu, Chun Huang, Mingzhe Cheng, Xiang Zhou, Lixian Department of Electrical and Computer Engineering University of Western Ontario LondonON Canada Department of Software Engineering GuangZhou Xinhua University GuangZhou China Department of Computer Science Michigan State University Michigan United States Department of Physics and Astronomy University College London London United Kingdom ChangJun High School International Department Changsha China

ISBN: (纸本)9781510664814

Adverse Drug Reaction (ADR) refer to harmful and irrelevant reactions that occur when normal dosage drugs are used to prevent, diagnose, treat diseases or regulate physiological functions. This definition excludes reactions caused by intentional or accidental overdose and inappropriate medication. In this paper, several models were measured and compared. The results demonstrated that base learners such as LR, SVM, RF, Adaboost, XGBoost may perform exceptionally well in some specific situations. On the other hand, if the precision of the outputs is emphasized, applying Stacking or even Multi-layer Stacking will be the most efficient tool. © 2023 SPIE.

关键词： Adaptive boosting

来源：评论

学校读者我要写书评

暂无评论

A Deep Learning Method to Translate Phonocardiogram (PCG) Signals to Electrocardiogram (ECG) Waveforms 13

A Deep Learning Method to Translate Phonocardiogram (PCG) Si...

引用

13th IEEE International Conference on CYBER Technology in Automation, Control, and Intelligent Systems, CYBER 2023

作者： Chen, Wei Li, Zhi Zhang, Guanglie College of Computer Science and Software Engineering Shenzhen University Shenzhen China City University of Hong Kong Department of Mechanical Engineering Hong Kong Hong Kong City University of Hong Kong Shenzhen Research Institute Shenzhen China

ISBN: (纸本)9798350315196

In clinical practice, an electrocardiogram (ECG), which analyzes the electrical activity of the human heart, is a common cardiovascular diagnostic test. The ECG data can be stored and diagnosed effectively. However, precise positioning of the ECG electrodes is necessary for a reliable ECG tracing. This has an impact on how ECG monitors are used to check on patients at home. On the other hand, the Phonocardiogram (PCG) is a standard method for quantifying the properties of heart sounds in order to monitor and assess cardiac activity. With the advancement of wearable sensors, it is now simpler to record the heart's sound. We proposed a deep learning approach to translating PCG signals to ECG waveforms. This technique can be utilized to monitor cardiovascular disease in consumer wearable device development. The proposed method is evaluated on a benchmark dataset of normal and anormal cardiac data. Experimental results show that our proposed approach is reasonable and effective. The generated ECG outputs can achieve a high accuracy, and the root mean squared error (RMSE) is 0.053. © 2023 IEEE.

关键词： Electrocardiograms

来源：评论

学校读者我要写书评

暂无评论

Deep Learning Based Intrusion Detection for IoT Networks

Deep Learning Based Intrusion Detection for IoT Networks

引用

Ubiquitous Computing and Intelligent Information Systems (ICUIS), International Conference on

作者： P. Ananthi K. Nirmaladevi Mothika G Department of Computer Technology-PG Kongu Engineering College Erode Tamilnadu Department of Computer Applications Madanapalle Institute of Technology &Science MSc Software Systems Kongu Engineering College Erode India

ISBN: (数字)9798331529635

ISBN: (纸本)9798331529642

Intrusion detection is crucial for securing IoT networks amid the rapid proliferation of devices, posing significant security challenges. This study offers a unique intrusion detection model for IoT deep learning methods used in networks include Convolutional Neural Networks (CNN), TabNet, and the Transient Search Optimization (TSO) method. The project starts by developing and evaluating distinct intrusion detection components. CNN, specializing in feature extraction, analyzes network traffic for potential intrusions. TabNet, known for interpretability and effectiveness with structured data, captures intricate relationships. The novel TSO algorithm improves the system's ability to detect temporary anomalies. Each component's performance is assessed individually, highlighting unique strengths. TabNet excels in structured IoT data pattern identification, surpassing CNN. The TSO algorithm strengthens temporary anomaly recognition, enhancing IoT network security. An ensemble technique combines CNN and TabNet predictions, reducing false positives and negatives to elevate overall detection accuracy. The ensemble, using both CNN and TabNet, achieves an impressive accuracy rate exceeding 99%, demonstrating significant gains in intrusion detection performance.

关键词： Deep learning Accuracy Intrusion detection Telecommunication traffic Prediction algorithms Ubiquitous computing Convolutional neural networks Internet of Things Transient analysis Optimization

来源：评论

学校读者我要写书评

暂无评论

An Efficient Blockchain-Based Healthcare System Using Artificial Intelligence

引用

computers, Materials & Continua 2022年第5期71卷 2721-2738页

作者： Aitizaz Ali Muhammad Fermi Pasha Ong Huey Fang Jehad Ali Mohammed A.Al.Zain Mehedi Masud Department of Software Systems and Cyber-Security School of ITMonash University47500Subang JayaMalaysia Department of Computer Engineering and Department of AI Convergence Network Ajou UniversitySuwon16499Korea Department of Information Technology College of Computers and Information TechnologyTaif University11099Taif21944Saudi Arabia Department of Computer Science College of Computers and Information TechnologyTaif University11099Taif21944Saudi Arabia

Personal health records and electronic health records are considered as the most sensitive information in the healthcare *** solutions have been provided for implementing the digital health system using blockchain,but there are several challenges,such as secure access control and privacy is one of the prominent ***,we propose a novel framework and implemented an attribute-based access control system using ***,we have also integrated artificial intelligence(AI)based approach to identify the behavior and activity for security *** current methods only focus on the related clinical records received from a medical ***,existing methods are too inflexible to resourcefully sustenance metadata changes.A secure patient data access framework is proposed in this research,integrating blockchain,trust chain,and blockchain methods to overcome these problems in the literature for sharing and accessing digital healthcare *** have used a neural network and classifier to categorize the user access to our proposed *** proposed scheme provides an intelligent and secure blockchain-based access control system in the digital healthcare *** results surpass the existing solutions by collecting attributes such as the number of transactions,number of nodes,transaction delay,block creation,and signature verification time.

关键词： Personal health records blockchain technology artificial intelligence access control attributes health care system

来源：评论

学校读者我要写书评

暂无评论

Accelerating the Task Activation and Data Communication for Dataflow Computing 22

Accelerating the Task Activation and Data Communication for ...

引用

Workshop Proceedings of the 51st International Conference on Parallel Processing

作者： Du Zheng Zhao Wenjie Wen Zhiwei Luo Qiuming Department of Computer Science and Technology Tsinghua University China College of Computer Science and Software Engineering Shenzhen University China

ISBN: (纸本)9781450394451

The hybrid dataflow/von-Neumann [1] architectures may differ in implementations but all follow similar principles: they harness the parallelism and data synchronization inherent to the dataflow model, yet maintain the programmability of the von-Neumann model. In this paper, we raise a new kind of hybrid dataflow/von-Neumann architectures, which contains TAU (Task Activated Unit) and SPM [9] (scratchpad memory) components, by which we can enhance parallel efficiency. We also implement the prototype design, integrated with peripheral devices and verify the whole system on FPGA. Finally, we deploy operating system on the hardware system and profile the performance. The experimental results show that the performance is improved by 3.07%∼10.32% under the random data flow graph, the performance of inter-core communication is improved by 4% and the hardware acceleration effect is achieved.

关键词： scratchpad memory RISC-V Dataflow/von-Neumann architecture parallel computation

来源：评论

学校读者我要写书评

暂无评论

2D-3D Feature Co-Embedding Network with Sparse Annotation for 3D Medical Image Segmentation

2D-3D Feature Co-Embedding Network with Sparse Annotation fo...

引用

2024 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2024

作者： Zhu, Mingzhu Guo, Shuyu Wang, Ziwei Jiao, Jianhang Bai, Tian Jilin University College of Computer Science and Technology Changchun China Jilin University Key Laboratory of Symbolic Computation and Knowledge Engineering Ministry of Education Changchun China Jilin University College of Software Changchun China The Second Norman Bethune Hospital Jilin University Department of Orthopedics Changchun China

ISBN: (纸本)9798350386226

Supervised methods on 3D medical image segmentation need large amounts of annotated data, but annotating is time-consuming. Also, existing 3D segmentation methods capture more global structural information but overlook local detailed features, which negatively impacts the segmentation of small tissues. In this paper, we propose a novel weakly-supervised 2D-3D Feature Co-Embedding Network (2D-3D CoENet) that includes 2D and 3D encoding layers, simultaneously extracting 2D local detailed and 3D global structural features. To reduce annotation costs, we use fewer labeled slices as ground truth and pseudo-labels are generated by 2D-3D CoENet for other slices. Additionally, multi-view learning is introduced to capture more 2D local detailed information, and a Multi-view Semantic Consistency loss (MSC loss) is proposed to constrain features from multiple perspectives. To further enhance the local detailed texture features, we propose an Edge Enhancement Module (EEM) in the 3D segmentation network to enhance the edge detail features. Our experimental results on the SKI10 dataset and OAI ZIB dataset demonstrate that our method outperforms the SOTA weakly-supervised segmentation methods. Moreover, our approach achieves results that are comparable to the fully-supervised upper bound results. © 2024 IEEE.

关键词： Medical imaging

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：