检索结果-内蒙古大学图书馆

26th International Conference on computer Supported Cooperative Work in Design, CSCWD 2023

作者： Wang, Baowei Yuan, Yi Li, Bin Dai, Changyu Wu, Yufeng Zheng, Weiqian Nanjing University of Information Science and Technology School of Computer and Software Nanjing China Ministry of Education Engineering Research Center of Digital Forensics Nanjing China Nanjing China

ISBN: (纸本)9798350331684

Recently, image crowdsourcing, a new trading mode, has been proposed to bridge the gap between the excess photos generated by intelligent devices and the great demand for images. However, traditional crowdsourcing methods often rely on centralized platforms, which risk data leakage and a single point of failure (SPOF). Moreover, due to the subjectivity of image quality assessment and the complexity of image data structure, image quality is difficult to control for traditional crowdsourcing frameworks without exposing data privacy. In this work, we propose a blockchain-based image crowdsourcing framework named QAIC to address these issues. Within the framework of QAIC, the transaction information is stored using a multichain structure, and the transaction process is implemented using smart contracts. We design an image selection and pricing mechanism for QAIC, where high-quality image sets can be spontaneously selected, and each image can be dynamically priced based on distortion degree and content relevance. Finally, to accurately obtain image quality, we design a dual output neural network model to evaluate the image quality, where a lightweight architecture is adopted, and piecewise outputs are designed to protect image privacy and reduce the on-chain computational cost Extensive analysis and experiments demonstrate that the quality of transaction data and reasonable pricing can be ensured using the QAIC without compromising image privacy. © 2023 IEEE.

关键词： Blockchain

来源：评论

学校读者我要写书评

暂无评论

Reinforcement Learning Application Testing Method based on Multi-attribute Fusion 9

Reinforcement Learning Application Testing Method based on M...

引用

9th International Conference on Dependable Systems and Their Applications, DSA 2022

作者： Cai, Lizhi Wang, Jin Chen, Mingang Wang, Jilong School of Information Science and Engineering East China University of Science and Technology Shanghai China Shanghai Development Center of Computer Software Technology Shanghai Key Laboratory of Computer Software Testing & Evaluating Shanghai China

ISBN: (数字)9781665488778

ISBN: (纸本)9781665488778

Reinforcement learning has been successfully applied to assess the reliability of applications, but the existing testing methods based on reinforcement learning have the problems of invalid interactive widgets and difficult training, resulting in low testing efficiency. In order to solve these problems, this paper proposes a lightweight application automation testing method of deep reinforcement learning based on multi-attribute fusion (MARTesting). First, the invalid widgets are removed by the difference operation of the attribute sets of the current and previous state, then the attributes of all widget elements on the page are abstracted into state as the input of the neural network, and a state is accurately determined by fusing the position and text information of page elements, finally combine the novelty of the state and the execution frequency of the action as a reward function. The experimental results on six open source applications show that the MARTesting method proposed in this paper has achieved significant improvements in code coverage and branch coverage compared with existing methods. © 2022 IEEE.

关键词： Reinforcement learning

来源：评论

学校读者我要写书评

暂无评论

Effective and Imperceptible Adversarial Textual Attack Via Multi-objectivization

引用

ACM Transactions on Evolutionary Learning and Optimization 2024年第3期4卷 1-23页

作者： Liu, Shengcai Lu, Ning Hong, Wenjing Qian, Chao Tang, Ke Department of Computer Science and Engineering Southern University of Science and Technology Shenzhen College of Engineering South Tower 1088 Xueyuan Avenue Guangdong Shenzhen518055 China Department of Computer Science and Engineering Hong Kong University of Science and Technology Hong Kong999077 Hong Kong National Engineering Laboratory for Big Data System Computing Technology Shenzhen University Shenzhen 10F Zhizhen Building Canghai Campus 3688 Nanhai Avenue Guangdong Shenzhen China National Key Laboratory for Novel Software Technology School of Artificial Intelligence Nanjing University Nanjing International College Area Xianlin Campus 163 Xianlin Avenue Jiangsu Nanjing China

The field of adversarial textual attack has significantly grown over the past few years, where the commonly considered objective is to craft adversarial examples (AEs) that can successfully fool the target model. However, the imperceptibility of attacks, which is also essential for practical attackers, is often left out by previous studies. In consequence, the crafted AEs tend to have obvious structural and semantic differences from the original human-written text, making them easily perceptible. In this work, we advocate leveraging multi-objectivization to address such an issue. Specifically, we reformulate the problem of crafting AEs as a multi-objective optimization problem, where the attack imperceptibility is considered as an auxiliary objective. Then, we propose a simple yet effective evolutionary algorithm, dubbed HydraText, to solve this problem. HydraText can be effectively applied to both score-based and decision-based attack settings. Exhaustive experiments involving 44,237 instances demonstrate that HydraText consistently achieves competitive attack success rates and better attack imperceptibility than the recently proposed attack approaches. A human evaluation study also shows that the AEs crafted by HydraText are more indistinguishable from human-written text. Finally, these AEs exhibit good transferability and can bring notable robustness improvement to the target model by adversarial training. © 2024 Copyright held by the owner/author(s).

关键词： Multiobjective optimization

来源：评论

学校读者我要写书评

暂无评论

On the Noise Robustness of In-Context Learning for Text Generation 38

On the Noise Robustness of In-Context Learning for Text Gene...

引用

38th Conference on Neural Information Processing Systems, NeurIPS 2024

作者： Gao, Hongfu Zhang, Feipeng Jiang, Wenyu Shu, Jun Zheng, Feng Wei, Hongxin Department of Statistics and Data Science Southern University of Science and Technology China School of Economics and Finance Xi'an Jiaotong University China National Key Laboratory for Novel Software Technology Nanjing University China School of Mathematics and Statistics Xi'an Jiaotong University China Department of Computer Science and Engineering Southern University of Science and Technology China

Large language models (LLMs) have shown impressive performance on downstream tasks by in-context learning (ICL), which heavily relies on the quality of demonstrations selected from a large set of annotated examples. Recent works claim that in-context learning is robust to noisy demonstrations in text classification. In this work, we show that, on text generation tasks, noisy annotations significantly hurt the performance of in-context learning. To circumvent the issue, we propose a simple and effective approach called Local Perplexity Ranking (LPR), which replaces the "noisy" candidates with their nearest neighbors that are more likely to be clean. Our method is motivated by analyzing the perplexity deviation caused by noisy labels and decomposing perplexity into inherent perplexity and matching perplexity. Our key idea behind LPR is thus to decouple the matching perplexity by performing the ranking among the neighbors in semantic space. Our approach can prevent the selected demonstrations from including mismatched input-label pairs while preserving the effectiveness of the original selection methods. Extensive experiments demonstrate the effectiveness of LPR, improving the EM score by up to 18.75 on common benchmarks with noisy annotations. Our code is available at https://***/ml-stat-Sustech/Local-Perplexity-Ranking. © 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

Design of 5G Vehicle-road-cloud Collaborative Architecture for Vehicle Platooning 8

Design of 5G Vehicle-road-cloud Collaborative Architecture f...

引用

8th Annual International Conference on Network and Information Systems for computers, ICNISC 2022

作者： Shi, Hongyan Qu, Xinran Ma, Jing Li, Jiajian School of Mechanical Engineering Dalian University of Technology Dut Dalian China School of Engineering Computer & Mathematical Sciences Auckland University of Technology Department of Computer Science & Software Engineering Auckland New Zealand

ISBN: (纸本)9781665453516

This paper presents an architecture design of vehicle-road-cloud collaboration based on a 5g communication network, which supports vehicle platooning and applies it in the case of vehicle formation crossing the border. The architecture includes roadside, cloud systems, 5g network, and infrastructure deployment connecting vehicles, as well as specifications for interfaces of automated vehicles (network communication interface specifications and application interface specifications). In addition, solutions and simple tests are proposed for the cross-border problems in the case of vehicle platooning crossing the border, mainly including interconnection delay between PLMNs, and low delay communication between leader and team members' vehicles and roadside infrastructure. © 2022 IEEE.

关键词： 5G mobile communication systems

来源：评论

学校读者我要写书评

暂无评论

Efficient and Effective Prompt Tuning via Prompt Decomposition and Compressed Outer Product

arXiv

引用

arXiv 2025年

作者： Lan, Pengxiang Xu, Haoyu Yang, Enneng Liang, Yuliang Guo, Guibing Zhao, Jianzhe Wang, Xingwei Software College Northeastern University China School of Computer Science and Engineering Northeastern University China

Prompt tuning (PT) offers a cost-effective alternative to fine-tuning large-scale pre-trained language models (PLMs), requiring only a few parameters in soft prompt tokens added before the input text. However, existing PT approaches face two significant issues: (i) They overlook intrinsic semantic associations between soft prompt tokens, leading to high discreteness and limited interactions, thus reducing the model’s comprehension and effectiveness in complex tasks. (ii) Due to the complexity of downstream tasks, long soft prompt is necessitated to improve performance, but prompt length correlates positively with memory usage and computational costs. Achieving high efficiency and performance remains an ongoing challenge. To address these issues, we propose a novel Low-pAraMeters Prompt Tuning (LAMP) method, which leverages prompt decomposition and compressed outer product. Specifically, the prompt decomposition module employs Truncated SVD to reduce training parameters and significantly lower the dimensionality of the soft prompt parameter space. It then utilizes a compressed outer product module to facilitate multiple interactions among prompt tokens, exploring their intrinsic associations to enhance knowledge representation. Finally, LAMP uses average pooling to reduce memory usage and training/inference time. Extensive experiments across six architectures and eight datasets demonstrate that LAMP outperforms state-of-the-art PT-based and LoRA-based methods in performance and efficiency. © 2025, CC BY.

关键词： Semantics

来源：评论

学校读者我要写书评

暂无评论

Research on Multi-Blockchain Electronic Archives Sharing Model

引用

computers, Materials & Continua 2023年第9期76卷 3921-3931页

作者： Fang Yu Wenbin Bi Ning Cao Jun Luo Diantang An Liqiang Ding Russell Higgs School of Information Engineering Qingdao Binhai UniversityQingdao266555China School of Computer and Software Dalian Neusoft University of InformationDalian116023China School of Internet of Things and Software Technology Wuxi Vocational College of Science and TechnologyWuxi214028China Archives Management Department QingdaoWest Coast New Area ArchivesQingdao266427China Editorial Research Department QingdaoWest Coast New Area ArchivesQingdao266427China School of Mathematics and Statistics University College DublinDublinIreland

The purpose of introducing blockchain into electronic archives sharing and utilization is to break the information barrier between electronic archives sharing departments by relying on technologies such as smart contract and asymmetric *** at the problem of dynamic permission management in common access control methods,a new access control method based on smart contract under blockchain is proposed,which improves the intelligence level under blockchain ***,the Internet attribute access control model based on smart contract is *** the dynamic access of heterogeneous devices,the management contract,permission judgment contract and access control contract are designed;Secondly,the access object credit evaluation algorithm based on particle swarm optimization radial basis function(PSO-RBF)neural network is used to dynamically generate the access node credit threshold combined with the access policy,so as to realize the intelligent access right management ***,combined with the abovemodels and algorithms,the workflow of electronic archives sharing and utilization model of multi blockchain is *** experimental results show that the timeconsuming of the process increases linearly with the number of continuous access to electronic archives blocks,and the secure access control of sharing and utilization is feasible,secure and effective.

关键词： Sharing and utilization of electronic archives dynamic permission management PSO-RBF neural network credit evaluation

来源：评论

学校读者我要写书评

暂无评论

A cloud-based framework for verifiable privacy-preserving spectrum auction

引用

High-Confidence Computing 2022年第1期2卷 47-53页

作者： Ruinian Li Tianyi Song Bo Mei Chunqiang Hu Wei Li Maya Larson Xiuzhen Cheng Rongfang Bie Department of Computer Science Bowling Green State UniversityBowling GreenOhioUSA Department of Computer Science Texas Christian UniversityTexasUSA Department of Software Engineering Chongqing UniversityChongqingChina Department of Computer Science Georgia State UniveristyAtlantaUSA Department of Computer Science the George Washington UniversityWashington DCUSA School of Computer Science and Technology Shandong UniversityQingdaoChina School of Information Science and Technology Beijing Normal UniversityBeijingChina

Spectrum auction is one of the most effective ways to achieve dynamic spectrum allocation in cognitive radio net-works,and it provides one effective way to manage the spectrum demands of IoT devices with limited *** spectrum auctions focus on protecting bidder privacy and achieving excellent social efficiency,but few tackles the verification of auction results that are controlled by the *** this paper,we propose a cloud-based framework for verifiable privacy-preserving spectrum *** framework adopts a modified AFGH re-encryption algorithm that achieves both bid privacy protection and auction results verification at the same *** cloud server helps to compute auction results based on homomorphic encryption,and an auctioneer decrypts the encrypted data from the server to obtain auction ***,the property of re-encryption makes it possible for any bidder to verify the auction results without compromising other bidders’privacy.

关键词： Homomorphic encryption Re-encryption Spectrum auction Privacy-preservation auction

来源：评论

学校读者我要写书评

暂无评论

Stress detection system for social media users

引用

Procedia computer science 2023年 216卷 672-681页

作者： Mochamad Rizky Febriansyah Nicholas Rezki Yunanda Derwin Suhartono Computer Science Department School of Computer Science Bina Nusantara University Jakarta 11480 Indonesia Software Engineering Program Computer Science Department School of Computer Science Bina Nusantara University Jakarta11480 Indonesia

Stress detection is a growing topic in the field of natural language processing. The study of stress detection for mental health prediction has been proven to benefit the development of recommender systems and automated mental health assessments in previous studies. Additionally, the widespread usage of social media has served as a potential data source for developing such models. Our research tried to detect whether the users of social media were under stress or not. We used a dataset from Dreaddit consisting of posts from one of the popular social media platforms, Reddit. We propose a machine learning model consisting of Support Vector Machine (SVM), Naïve Bayes, Decision Tree, Random Forest, Bag of Words, and Term Frequency – inverse document frequency (TF-IDF) for stress detection. The final evaluation of the model achieved an 80.00% F-1 Score and 75.00% accuracy, and both were scored by SVM.

关键词： Machine Learning Stress Detection Social Media Natural Language Processing

来源：评论

学校读者我要写书评

暂无评论

A Novel Redundant Service Caching and Task Offloading Method in Mobile Edge Computing 31st

A Novel Redundant Service Caching and Task Offloading Metho...

引用

31st International Conference on Web Services, ICWS 2024, Held as Part of the Services Conference Federation, SCF 2024

作者： Ouyang, Zhaobin Xia, Yunni Li, Jingpeng Feng, Jiafeng Yu, Yang Zhang, Ke Xu, Xifeng Ma, Yong Chen, Peng Li, Xiaobo School of Computer Chongqing University Chongqing400030 China China Huadian Corporation Guangdong Branch Guangdong China Nanjing Nanzi Huadun Digital Technology Co. Ltd. Nanjing China School of Computer Science and Technology Beijing Institute of Technology Beijing100083 China School of Computer and Information Engineering Jiangxi Normal University Nanchang330000 China School of Computer and Software Engineering Xihua University Chengdu China Chongqing Animal Husbandry Techniques Extension Center Chongqing China

ISBN: (纸本)9783031770715

Edge caching is a promising technologies, which enhances content delivery and reduces service latency by caching potentially interested content and services on edge nodes near users. Due to constraints such as limited edge server capacities and unstable communications in a highly dynamic MEC environment, caching and content delivery are prone to failures and inconsistencies. Consequently, redundant caching mechanisms are in high need as counter measures in this paper, we proposes a novel redundant service caching and task offloading Decision (FT-STD) method. Specifically, we utilize the primary backup (PB) approach for offloading fault tolerance and employ a learning-based algorithm for yielding redundant service caching schedules in MEC. Simulation results demonstrate thatclearly outperforms its peers across multiple performance metrics. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

关键词： Mobile edge computing

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：