检索结果-内蒙古大学图书馆

KD-Crowd:a knowledge distillation framework for learning from crowds

Frontiers of computer science 2025年第1期19卷 119-130页

作者： Shaoyuan LI Yuxiang ZHENG Ye SHI Shengjun HUANG Songcan CHEN College of Computer Science and Technology Nanjing University of Aeronautics and AstronauticsMIIT Key Laboratory of Pattern Analysis and Machine IntelligenceNanjing 211106China

Recently, crowdsourcing has established itself as an efficient labeling solution by distributing tasks to crowd workers. As the workers can make mistakes with diverse expertise, one core learning task is to estimate each worker’s expertise, and aggregate over them to infer the latent true labels. In this paper, we show that as one of the major research directions, the noise transition matrix based worker expertise modeling methods commonly overfit the annotation noise, either due to the oversimplified noise assumption or inaccurate estimation. To solve this problem, we propose a knowledge distillation framework (KD-Crowd) by combining the complementary strength of noise-model-free robust learning techniques and transition matrix based worker expertise modeling. The framework consists of two stages: in Stage 1, a noise-model-free robust student model is trained by treating the prediction of a transition matrix based crowdsourcing teacher model as noisy labels, aiming at correcting the teacher’s mistakes and obtaining better true label predictions;in Stage 2, we switch their roles, retraining a better crowdsourcing model using the crowds’ annotations supervised by the refined true label predictions given by Stage 1. Additionally, we propose one f-mutual information gain (MIG^(f)) based knowledge distillation loss, which finds the maximum information intersection between the student’s and teacher’s prediction. We show in experiments that MIG^(f) achieves obvious improvements compared to the regular KL divergence knowledge distillation loss, which tends to force the student to memorize all information of the teacher’s prediction, including its errors. We conduct extensive experiments showing that, as a universal framework, KD-Crowd substantially improves previous crowdsourcing methods on true label prediction and worker expertise estimation.

关键词： crowdsourcing label noise worker expertise knowledge distillation robust learning

来源：评论

学校读者我要写书评

暂无评论

SmartEagleEye:A Cloud-Oriented Webshell Detection System Based on Dynamic Gray-Box and Deep Learning

引用

Tsinghua science and technology 2024年第3期29卷 766-783页

作者： Xin Liu Yingli Zhang Qingchen Yu Jiajun Min Jun Shen Rui Zhou Qingguo Zhou School of Information Science and Engineering Lanzhou UniversityLanzhou 730000China College of Computer Science and Technology Zhejiang UniversityHangzhou 310058China School of Computing and Information Technology University of WollongongWollongong 2500Australia

Compared with traditional environments,the cloud environment exposes online services to additional vulnerabilities and threats of cyber attacks,and the cyber security of cloud platforms is becoming increasingly prominent.A piece of code,known as a Webshell,is usually uploaded to the target servers to achieve multiple *** Webshell attacks has become a hot spot in current ***,the traditional Webshell detectors are not built for the cloud,making it highly difficult to play a defensive role in the cloud ***,a Webshell detection system based on deep learning that is successfully applied in various scenarios,is proposed in this *** system contains two important components:gray-box and neural network *** gray-box analyzer defines a series of rules and algorithms for extracting static and dynamic behaviors from the code to make the decision *** neural network analyzer transforms suspicious code into Operation Code(OPCODE)sequences,turning the detection task into a classification *** experiment results show that SmartEagleEye achieves an encouraging high detection rate and an acceptable false-positive rate,which indicate its capability to provide good protection for the cloud environment.

关键词： Webshell detection cloud web security deep learning

来源：评论

学校读者我要写书评

暂无评论

Abnormal event detection in surveillance videos through LSTM auto-encoding and local minima assistance

引用

Discover Internet of Things 2025年第1期5卷 1-25页

作者： Sengonul, Erkan Samet, Refik Abu Al-Haija, Qasem Alqahtani, Ali Alsemmeari, Rayan A. Alghamdi, Bandar Alturki, Badraddin Alsulami, Abdulaziz A. Department of Computer Engineering Ankara University Ankara06100 Turkey Department of Cybersecurity Faculty of Computer & Information Technology Jordan University of Science and Technology PO Box 3030 Irbid22110 Jordan Department of Networks & Communications Engineering College of Computer Science & Information Systems Najran University Najran61441 Saudi Arabia Department of Information Technology Faculty of Computing and Information Technology King Abdulaziz University Jeddah21589 Saudi Arabia Department of Information Systems Faculty of Computing and Information Technology King Abdulaziz University Jeddah21589 Saudi Arabia

Abnormal event detection in video surveillance is critical for security, traffic management, and industrial monitoring applications. This paper introduces an innovative methodology for anomaly detection in video data, encompassing three primary stages: preprocessing, feature learning, and anomaly detection. We employ background subtraction and noise reduction during preprocessing to refine the data. The feature-learning stage involves training an LSTM autoencoder to capture the essential features of normal video sequences. For anomaly detection, we map video data to a lower-dimensional space (latent code) and compare it against the distribution of codes from normal sequences. We determine regularity scores and identify local minima points exceeding a specified threshold while scrutinizing shadows between adjacent maxima to confirm and pinpoint anomalies. When tested on the CUHK Avenue and UMN datasets, our methodology demonstrated performance with AUCs of 93.8% and 94.1%, respectively, outperforming several baseline models. Our results show the high precision of our method that can detect anomalies, highlighting its potential advantages that it can achieve for enhancing systems of surveillance. © The Author(s) 2025.

关键词： Noise abatement

来源：评论

学校读者我要写书评

暂无评论

Pyramid-VAE-GAN:Transferring hierarchical latent variables for image inpainting

引用

Computational Visual Media 2023年第4期9卷 827-841页

作者： Huiyuan Tian Li Zhang Shijian Li Min Yao Gang Pan College of Computer Science and Technology Zhejiang UniversityHangzhou 310027China Advanced Technology Research Institute Zhejiang UniversityHangzhou 310027China

Significant progress has been made in image inpainting methods in recent ***,they are incapable of producing inpainting results with reasonable structures,rich detail,and sharpness at the same *** this paper,we propose the Pyramid-VAE-GAN network for image inpainting to address this *** network is built on a variational autoencoder(VAE)backbone that encodes high-level latent variables to represent complicated high-dimensional prior distributions of *** prior assists in reconstructing reasonable structures when *** also adopt a pyramid structure in our model to maintain rich detail in low-level latent *** avoid the usual incompatibility of requiring both reasonable structures and rich detail,we propose a novel cross-layer latent variable transfer *** transfers information about long-range structures contained in high-level latent variables to low-level latent variables representing more detailed *** further use adversarial training to select the most reasonable results and to improve the sharpness of the *** experimental results on multiple datasets demonstrate the superiority of our *** code is available at https://***/thy960112/Pyramid-VAE-GAN.

关键词： image inpainting variational autoencoder(VAE) latent variable transfer(LTN) pyramid structure generative model

来源：评论

学校读者我要写书评

暂无评论

A State-Migration Particle Swarm Optimizer for Adaptive Latent Factor Analysis of High-Dimensional and Incomplete Data

引用

IEEE/CAA Journal of Automatica Sinica 2024年第11期11卷 2220-2235页

作者： Jiufang Chen Kechen Liu Xin Luo Ye Yuan Khaled Sedraoui Yusuf Al-Turki MengChu Zhou IEEE the College of Computer Science and Technology Chongqing University of Posts and Telecommunications the Department of Computer Science Columbia University the College of Computer and Information Science Southwest University the Department of Electrical and Computer Engineering Faculty of Engineeringand also with K.A.CARE Energy Research and Innovation CenterKing Abdulaziz University the Center of Research Excellence in Renewable Energy and Power Systems the Department of Electrical and Computer EngineeringFaculty of Engineeringand also with K.A.CARE Energy Research and Innovation CenterKing Abdulaziz University the Department of Electrical and Computer Engineering New Jersey Institute of Technology

High-dimensional and incomplete(HDI) matrices are primarily generated in all kinds of big-data-related practical applications. A latent factor analysis(LFA) model is capable of conducting efficient representation learning to an HDI matrix,whose hyper-parameter adaptation can be implemented through a particle swarm optimizer(PSO) to meet scalable ***, conventional PSO is limited by its premature issues,which leads to the accuracy loss of a resultant LFA model. To address this thorny issue, this study merges the information of each particle's state migration into its evolution process following the principle of a generalized momentum method for improving its search ability, thereby building a state-migration particle swarm optimizer(SPSO), whose theoretical convergence is rigorously proved in this study. It is then incorporated into an LFA model for implementing efficient hyper-parameter adaptation without accuracy loss. Experiments on six HDI matrices indicate that an SPSO-incorporated LFA model outperforms state-of-the-art LFA models in terms of prediction accuracy for missing data of an HDI matrix with competitive computational ***, SPSO's use ensures efficient and reliable hyper-parameter adaptation in an LFA model, thus ensuring practicality and accurate representation learning for HDI matrices.

关键词： Data science generalized momentum high-dimensional and incomplete(HDI)data hyper-parameter adaptation latent factor analysis(LFA) particle swarm optimization(PSO)

来源：评论

学校读者我要写书评

暂无评论

Deployment method of wireless sensor networks based on MaOEA/P-GM algorithm

引用

Soft Computing 2024年 1-10页

作者： Wang, Jun Luo, Dongxu Chen, Weiru Peng, Funan Li, Zhiwu College of Computer Science and Technology Shenyang University of Chemical Technology Shenyang110142 China Key Laboratory of Industrial Intelligence Technology on Chemical Process Shenyang University of Chemical Technology Shenyang110142 China College of System Engineering Institute Macao University of Science and Technology 00853 China

At present, wireless sensor networks are developing rapidly, but they will also face many challenges. For example, in the deployment problem, many problems such as cost, coverage, connectivity, energy, network life cycle and so on need to be considered, while the traditional multi-objective algorithm does not perform well in dealing with many-objective problems. How to quickly find high-quality solutions is the focus of research today. This paper proposes a deployment method for wireless sensor networks based on MaOEA/P-GM algorithm. This method improves the mutation strategy of the algorithm according to the characteristics of wireless sensor network deployment based on the many-objective algorithm MaOEA/P. Experiments show that using this algorithm in a many-objective environment can quickly get a high-quality solution set with fewer nodes, high coverage, good connectivity, and balanced energy consumption. © The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature 2024.

关键词： Life cycle

来源：评论

学校读者我要写书评

暂无评论

Generative adversarial networks based motion learning towards robotic calligraphy synthesis

引用

CAAI Transactions on Intelligence technology 2024年第2期9卷 452-466页

作者： Xiaoming Wang Yilong Yang Weiru Wang Yuanhua Zhou Yongfeng Yin Zhiguo Gong Department of Computer and Information Science University of MacaoMacaoChina School of Software Beihang UniversityBeijingChina Department of Computer Science and Technology Faculty of Information TechnologyBeijing University of TechnologyBeijingChina School of Foreign Languages Guangzhou Huashang CollegeGuangzhouChina

Robot calligraphy visually reflects the motion capability of robotic *** traditional researches mainly focus on image generation and the writing of simple calligraphic strokes or characters,this article presents a generative adversarial network(GAN)-based motion learning method for robotic calligraphy synthesis(Gan2CS)that can enhance the efficiency in writing complex calligraphy words and reproducing classic calligraphy *** key technologies in the proposed approach include:(1)adopting the GAN to learn the motion parameters from the robot writing operation;(2)converting the learnt motion data into the style font and realising the transition from static calligraphy images to dynamic writing demonstration;(3)reproducing high-precision calligraphy works by synthesising the writing motion data *** this study,the motion trajectories of sample calligraphy images are firstly extracted and converted into the robot *** robot performs the writing with motion planning,and the writing motion parameters of calligraphy strokes are learnt with *** the motion data of basic strokes is synthesised based on the hierarchical process of‘stroke-radicalpart-character’.And the robot re-writes the synthesised characters whose similarity with the original calligraphy characters is *** calligraphy characters have been tested in the experiments for method validation and the results validated that the robot can actualise the robotic calligraphy synthesis of writing motion data with GAN.

关键词： calligraphy synthesis generative adversarial networks Motion learning robot writing

来源：评论

学校读者我要写书评

暂无评论

Point cloud completion via structured feature maps using a feedback network

引用

Computational Visual Media 2023年第1期9卷 71-85页

作者： Zejia Su Haibin Huang Chongyang Ma Hui Huang Ruizhen Hu College of Computer Science&Software Engineering Shenzhen UniversityShenzhenChina Kuaishou Technology China

In this paper,we tackle the challenging problem of point cloud completion from the perspective of feature *** key observation is that to recover the underlying structures as well as surface details,given partial input,a fundamental component is a good feature representation that can capture both global structure and local geometric *** accordingly first propose FSNet,a feature structuring module that can adaptively aggregate point-wise features into a 2D structured feature map by learning multiple latent patterns from local *** then integrate FSNet into a coarse-to-fine pipeline for point cloud ***,a 2D convolutional neural network is adopted to decode feature maps from FSNet into a coarse and complete point ***,a point cloud upsampling network is used to generate a dense point cloud from the partial input and the coarse intermediate *** efficiently exploit local structures and enhance point distribution uniformity,we propose IFNet,a point upsampling module with a self-correction mechanism that can progressively refine details of the generated dense point *** have conducted qualitative and quantitative experiments on ShapeNet,MVP,and KITTI datasets,which demonstrate that our method outperforms stateof-the-art point cloud completion approaches.

关键词： 3D point clouds shape completion geometry processing deep learning

来源：评论

学校读者我要写书评

暂无评论

All optical on-chip nonlinear activation functions for multi-wavelength computation

引用

Optics Express 2025年第10期33卷 20852-20862页

作者： Xia, Gongyu Liu, Jiacheng Zhu, Pingyu Hong, Qilin Xu, Ping Zhu, Zhihong College of Advanced Interdisciplinary Studies National University of Defense Technology Changsha410073 China College of Computer Science and Technology National University of Defense Technology Changsha410073 China

This work proposes an all-optical on-chip nonlinear activation function unit for multiwavelength computing applications. Our proposed unit utilizes the unique properties of silicon, specifically the two-photon absorption and thermo-optic effects, to achieve nonlinearity essential for optical neural networks. We designed the parameters of the corresponding unit structure through simulation, fabricated and tested the actual structure experimentally, and obtained a CELU-like response curve. In practical classification tasks, we evaluated its performance in various classification tasks through numerical simulation and achieved a recognition accuracy of 98.84% on the MNIST dataset. This work addresses a critical gap in multi-wavelength systems and contributes to the development of more powerful and efficient optical computing platforms. © 2025 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement.

关键词： Machine learning Neural networks Numerical simulation Optical computing Optical neural systems Optical signals

来源：评论

学校读者我要写书评

暂无评论

Online Information-Based Product Evolution Course Mining and Prediction

引用

International Journal of Information technology and Decision Making 2024年第2期23卷 599-627页

作者： Chen, Yuh-Jen Chen, Yuh-Min Department of Accounting and Information Systems College of Business Intelligence National Kaohsiung University of Science and Technology Kaohsiung City Taiwan Institute of Manufacturing Information and Systems College of Electrical Engineering and Computer Science National Cheng Kung University Tainan City Taiwan

In today’s fiercely competitive business environment, product design and development play critical roles in an enterprise’s success;therefore, consumer demand must be understood. Enterprises used to understand their consumers’ demands through time-consuming questionnaire surveys and statistical analyses. As the Internet and the popularity of virtual communities have grown, more consumers are leaving comments about their perceptions of the appeal of products on online social media platforms, thus enabling enterprises to more objectively understand consumers’ product preferences and demands. Therefore, determining how to effectively assist enterprises in analyzing valuable information beneficial to product design and development that can be gleaned from the large amount of social media data available is critical to promoting an enterprise’s competitive advantage in the product market. However, previous studies have primarily focused on understanding consumer viewpoints of products through review articles or electronic word-of-mouth (eWOM;it is a common channel for spreading product appraisals) from online social media. For these review articles and eWOM that can imply consumer demand for product features, there were no relevant studies focused on analyzing consumer demand by using both online review articles and eWOM for product feature evolution. Therefore, for new product features, this study developed a mechanism for product evolution course mining from product-related review articles (such as product functions or specifications) and eWOM on online social media to realize the prediction of future product features or specifications, and then assist enterprises in rapidly and accurately grasping product development trends to effectively discern key reference information for product design and development. The study was achieved by (1) designing a process for online information-based product evolution course mining and prediction, (ii) developing techniques related to

关键词： Forecasting

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：