检索结果-内蒙古大学图书馆

arXiv 2023年

作者： Jiang, Shuyu Tang, Wenyi Chen, Xingshu Tang, Rui Wang, Haizhou Wang, Wenxian School of Cyber Science and Engineering Sichuan University Chengdu China Cyber Science Research Institute Sichuan University Chengdu China Key Laboratory of Data Protection and Intelligent Management Sichuan University Ministry of Education Chengdu China

The proliferation of hate speech (HS) on social media poses a serious threat to societal security. Automatic counter narrative (CN) generation, as an active strategy for HS intervention, has garnered increasing attention in recent years. Existing methods for automatically generating CNs mainly rely on re-training or fine-tuning pre-trained language models (PLMs) on human-curated CN corpora. Unfortunately, the annotation speed of CN corpora cannot keep up with the growth of HS targets, while generating specific and effective CNs for unseen targets remains a significant challenge for the model. To tackle this issue, we propose Retrieval-Augmented Zero-shot Generation (ReZG) to generate CNs with high-specificity for unseen targets. Specifically, we propose a multi-dimensional hierarchical retrieval method that integrates stance, semantics, and fitness, extending the retrieval metric from single dimension to multiple dimensions suitable for the knowledge that refutes HS. Then, we implement an energy-based constrained decoding mechanism that enables PLMs to use differentiable knowledge preservation, countering, and fluency constraint functions instead of in-target CNs as control signals for generation, thereby achieving zero-shot CN generation. With the above techniques, ReZG can integrate external knowledge flexibly and improve the specificity of CNs. Experimental results show that ReZG exhibits stronger generalization capabilities and outperforms strong baselines with significant improvements of 2.0%+ in the relevance and 4.5%+ in the countering success rate metrics. © 2023, CC BY-NC-ND.

关键词： Semantics

来源：评论

学校读者我要写书评

暂无评论

GPTCN: Gated Parallel Transformer Convolutional Networks for Downstream-Task User Representation Learning on App Usage

GPTCN: Gated Parallel Transformer Convolutional Networks for...

引用

International Conference on Acoustics, Speech, and Signal Processing (ICASSP)

作者： Yingjie Sun Fanrui Zeng Jiamin Xiao Yuxiao Deng Yifan Ding Yizhou Li School of Cyber Science and Engineering Sichuan University Chengdu China Key Laboratory of Data Protection and Intelligent Management Ministry of Education Sichuan University Shenzhen Hexun Huagu Information Technology Co. Ltd Shanghai China

With the development of mobile applications into a part of modern life, the user usage behavior data of mobile applications can well reflect the attribute characteristics of users. For many downstream applications, including advertising, recommendations provide effective support. To provide users with customized services and optimize user experience, industry and scholars have been exploring feasible solutions. However, automatic user modeling based on mobile app usage faces unique challenges, including (1) poor generalization performance of modeling with a single downstream task, (2) uneven distribution of user behavior over time, and severe sparsity in many long-tail apps. In this paper, we propose a custom model (GPTCN) by which we overcome these challenges and optimize on top of it, achieving the goal of reducing manual effort and improving performance. Experimental results show that GPTCN outperforms the state-of-the-art general vector generation models in ACC, providing a more comprehensive, robust and general user representation model.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Prompt Packer: Deceiving LLMs through Compositional Instruction with Hidden Attacks

arXiv

引用

arXiv 2023年

作者： Jiang, Shuyu Chen, Xingshu Tang, Rui School of Cyber Science and Engineering Sichuan University Chengdu China Cyber Science Research Institute Sichuan University Chengdu China Key Laboratory of Data Protection and Intelligent Management Ministry of Education Sichuan University Chengdu China

Recently, Large language models (LLMs) with powerful general capabilities have been increasingly integrated into various Web applications, while undergoing alignment training to ensure that the generated content aligns with user intent and ethics. Unfortunately, they remain the risk of generating harmful content like hate speech and criminal activities in practical applications. Current approaches primarily rely on detecting, collecting, and training against harmful prompts to prevent such risks. However, they typically focused on the "superficial" harmful prompts with a solitary intent, ignoring composite attack instructions with multiple intentions that can easily elicit harmful content in real-world scenarios. In this paper, we introduce an innovative technique for obfuscating harmful instructions: Compositional Instruction Attacks (CIA), which refers to attacking by combination and encapsulation of multiple instructions. CIA hides harmful prompts within instructions of harmless intentions, making it impossible for the model to identify underlying malicious intentions. Furthermore, we implement two transformation methods, known as T-CIA and W-CIA, to automatically disguise harmful instructions as talking or writing tasks, making them appear harmless to LLMs. We evaluated CIA on GPT-4, ChatGPT, and ChatGLM2 with two safety assessment datasets and two harmful prompt datasets. It achieves an attack success rate of 95%+ on safety assessment datasets, and 83%+ for GPT-4, 91%+ for ChatGPT (gpt-3.5-turbo backed) and ChatGLM2-6B on harmful prompt datasets. Our approach reveals the vulnerability of LLMs to such compositional instruction attacks that harbor underlying harmful intentions, contributing significantly to LLM security development. Warning: this paper may contain offensive or upsetting content!. Copyright © 2023, The Authors. All rights reserved.

关键词： Computational linguistics

来源：评论

学校读者我要写书评

暂无评论

Analysis of conditional connectivity based on two lanes for VANETs 12th

Analysis of conditional connectivity based on two lanes for ...

引用

12th International Conference on Communications and Networking in China, CHINACOM 2017

作者： Pan, Bin Wu, Hao State Key Laboratory of Rail Traffic Control and Safety Beijing Jiaotong University Beijing100044 China School of Electronic and Information Engineering Beijing Jiaotong University Beijing100044 China Beijing Key Laboratory of Intelligent Traffic Data Safety and Privacy Protection Technology Beijing Jiaotong University Beijing100044 China Beijing Engineering Research Center of High-Speed Railway Broadband Mobile Communications Beijing Jiaotong University Beijing100044 China

ISBN: (纸本)9783319781389

Conditional connectivity is important for the design of the upper layer communication protocol and the network deployment in different scenarios. This paper analyzes the performance of conditional connectivity with the network topology change by establishing the model of inter-vehicle communication. The paper aims at the two-lane highway scenario for vehicular ad hoc networks (VANETs), considering the factors of communication range, vehicle flow characteristics and moving speed. The main conditional connectivity performance index is the conditional connectivity probability. Based on the simulation, the correctness of the theoretical analysis is verified. We also make an explanation of the simulation results. © ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2018.

关键词： Vehicular ad hoc networks

来源：评论

学校读者我要写书评

暂无评论

Delivery delay analysis for large-size multimedia messages in V2I/V2V networks 7

Delivery delay analysis for large-size multimedia messages i...

引用

7th IET International Conference on Wireless, Mobile and Multimedia Networks, ICWMMN 2017

作者： Pan, Bin Wu, Hao Ding, Hongwei State Key Laboratory of Rail Traffic Control and Safety Beijing Jiaotong University Beijing100044 China School of Electronic and Information Engineering Beijing Jiaotong University Beijing100044 China Beijing Key laboratory of intelligent traffic data safety and privacy protection technology Beijing Jiaotong University Beijing100044 China Beijing Engineering Research Center of High-speed Railway Broadband Mobile Communications Beijing Jiaotong University Beijing100044 China

ISBN: (纸本)9781785615153

This paper analyzes the large-size messages delivery delay in hybrid Vehicle-to-Infrastructure (V2I) and Vehicle-to-Vehicle (V2V) networks. An analytical model is developed to describe the relationship between the average delivery delay and the distance between neighbor Roadside Units (RSUs) deployed along a road. The effect of message size has been also analysed. The derived model considers a straight highway scenario where RSUs are deployed at the same mter-RSU distance without any direct connection and vehicles are sparsely distributed on the road. We also take into account the dissemination strategy, aiming to minimize the delivery delay. The effectiveness of the derived analytical model is verified through simulation results. The analytical model can also provide important insights for the deployment of RSUs and related reference for the onboard multimedia messages selected to share with others. © Institution of Engineering and Technology. All rights reserved.

关键词： Vehicular ad hoc networks

来源：评论

学校读者我要写书评

暂无评论

Dynamic Topic Model for Tracking Topic Evolution and Measuring Popularity of Scientific Literature

Dynamic Topic Model for Tracking Topic Evolution and Measuri...

引用

IEEE International Conference on data Science in Cyberspace (DSC)

作者： Yezheng Liu Jicheng Wang Yang Qian Yuanchun Jiang Jianshan Sun Yidong Chai School of Management Hefei University of Technology National Engineering Laboratory for Big Data Distribution and Exchange Technologies Key Laboratory of Process Optimization and Intelligent Decision Making Ministry of Education

ISBN: (纸本)9781665418164

With the progress of science and technology, a large number of scientific papers are published every year. Faced with such large data, identifying high-value research and hot research directions has become an interesting and important task. Previous studies focus less on the popularity measurement of research papers. In this paper, we use a dynamic topic model (DTM) for tracking the topic evolution trend and measuring the popularity of research papers based on the topic hotness. To test our model effectiveness, we conduct experiments on 5,964 research papers in the category of management. Our results show that DTM can help to monitor topic evolution and measuring popularity for scientific literature.

关键词： Current measurement Conferences Cyberspace data science Market research data models Time factors

来源：评论

学校读者我要写书评

暂无评论

A Low-cost, High-impact Node Injection Approach for Attacking Social Network Alignment

arXiv

引用

arXiv 2023年

作者： Jiang, Shuyu Qiu, Yunxiang Mo, Xian Tang, Rui Wang, Wei School of Cyber Science and Engineering Sichuan University Chengdu610065 China The Key Laboratory of Data Protection and Intelligent Management Ministry of Education Sichuan University Chengdu610065 China Huaxin Consulting and Designing Institute Co. Ltd Hangzhou310052 China The School of Information Engineering Ningxia University Yinchuan750021 China The School of Cyber Science and Engineering The Key Laboratory of Data Protection and Intelligent Management Ministry of Education Sichuan University Chengdu610065 China The School of Public Health Chongqing Medical University Chongqing401331 China

Social network alignment (SNA) holds significant importance for various downstream applications, prompting numerous professionals to develop and share SNA tools. Unfortunately, these tools can be exploited by malicious actors to integrate sensitive user information, posing cybersecurity risks. While many researchers have explored attacking SNA (ASNA) through a network modification attack way, practical feasibility remains a challenge. This paper introduces a novel approach, the node injection attack. To overcome the problem of modeling and solving within a limited time and balancing costs and benefits, we propose a low-cost, high-impact node injection attack via dynamic programming (DPNIA) framework. DPNIA models ASNA as a problem of maximizing the number of confirmed incorrect correspondent node pairs who have a greater similarity scores than the pairs between existing nodes, making ASNA solvable. Meanwhile, it employs a cross-network evaluation method to identify node vulnerability, facilitating a progressive attack from easy to difficult. Additionally, it utilizes an optimal injection strategy searching method, based on dynamic programming, to determine which links should be added between injected nodes and existing nodes, thereby achieving a high impact for attack effectiveness at a low cost. Experiments on four real-world datasets consistently demonstrate that DPNIA consistently and significantly outperforms various attack baselines. Copyright © 2023, The Authors. All rights reserved.

关键词： Dynamic programming

来源：评论

学校读者我要写书评

暂无评论

Global, regional, and national burden of HIV-negative tuberculosis, 1990-2021: findings from the Global Burden of Disease Study 2021

引用

Infectious Diseases of Poverty 2024年第4期13卷 101-102页

作者： Shun-Xian Zhang Feng-Yu Miao Jian Yang Wen-Ting Zhou Shan Lv Fan-Na Wei Yu Wang Xiao-Jie Hu Ping Yin Pei-Yong Zheng Ming Yang Mei-TiWang Xin-Yu Feng Lei Duan Guo-Bing Yang Ji-Chun Wang Zhen-Hui Lu Longhua Hospital Shanghai University of Traditional Chinese Medicine Shanghai 200032 China National Institute of Parasitic Diseases at Chinese Center for Disease Control and Prevention (Chinese Center for Tropical Diseases Research) NHC Key Laboratory of Parasite and Vector Biology WHO Collaborating Centre for Tropical Diseases National Center for International Research On Tropical Diseases National Key Laboratory of Intelligent Tracking and Forecasting for Infectious Diseases Shanghai 200025 China School of Global Health Chinese Center for Tropical Diseases Research-Shanghai Jiao Tong University School of Medicine Shanghai 200025 China Beijing Municipal Health Big Data and Policy Research Center Beijing Institute of Hospital Management Beijing 101100 China Department of Science and Technology Chinese Center for Disease Control and Prevention National Key Laboratory of Intelligent Tracking and Forecasting for Infectious Diseases Beijing 102206 China National Health Commission (NHC) Key Laboratory of Biosafety National Institute for Viral Disease Control and Prevention Chinese Center for Disease Control and Prevention National Key Laboratory of Intelligent Tracking and Forecasting for Infectious Diseases Beijing 102206 China Shanghai Mental Health Center Shanghai Jiao Tong University School of Medicine Shanghai 200032 China Gansu Provincial Center for Disease Control and Prevention Lanzhou 730000 China

Background Tuberculosis(TB)is a major infectious disease with significant public health *** widespread transmission,prolonged treatment duration,notable side effects,and high mortality rate pose severe *** study examines the epidemiological characteristics of TB globally and across major regions,providing a scientific basis for enhancing TB prevention and control measures *** The ecological study used data from the Global Burden of Disease(GBD)Study *** assessed new incidence cases,deaths,disability-adjusted life years(DALYs),and trends in age-standardized incidence rates(ASIRs),mortality rates(ASMRs),and DALY rates for drug-susceptible tuberculosis(DS-TB),multidrug-resistant tuberculosis(MDR-TB),and extensively drug-resistant tuberculosis(XDR-TB)from 1990 to 2021.A Bayesian age-period-cohort model was applied to project ASIR and *** In 2021,the global ASIR for all HIV-negative TB was 103.00 per 100,000 population 95%uncertainty interval(UI):92.21,114.91 per 100,000 populationl,declining by 0.40%(95%UI:-0.43,-0.38%)compared to *** global ASMR was 13.96 per 100,000 population(95%UI:12.61,15.72 per 100,000 population),with a decline of 0.44%(95%UI:-0.61,-0.23%)since *** global age-standardized DALY rate for HIV-negative TB was 580.26 per 100,000 population(95%UI:522.37,649.82 per 100,000 population),showing a decrease of 0.65%(95%UI:-0.69,-0.57 per 100,000 population)from *** global ASIR of MDR-TB has not decreased since 2015,instead,it has shown a slow upward trend in recent *** ASIR of XDR-TB has exhibited significant increase in the past 30 *** projections indicate MDR-TB and XDR-TB are expected to see signifcant increases in both ASIR and ASMR from 2022 to 2035,highlighting the growing challenge of drug-resistant *** This study found that the ASIR of MDR-TB and XDR-TB has shown an upward trend in recent *** reduce the TB burden,it is essential to enhance health infrastru

关键词： Tuberculosis Epidemiology Global burden of disease Sociodemographic Index

来源：评论

学校读者我要写书评

暂无评论

S2CCT: Self-Supervised Collaborative CNN-Transformer for Few-shot Medical Image Segmentation

S2CCT: Self-Supervised Collaborative CNN-Transformer for Few...

引用

2024 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2024

作者： Zhou, Rongzhou Shu, Ziqi Xie, Weixing Yao, Junfeng Hong, Qingqi Xiamen University National Institute for Data Science in Health and Medicine China Xiamen University Center for Digital Media Computing School of Film School of Informatics Xiamen361005 China Key Laboratory of Digital Protection and Intelligent Processing of Intangible Cultural Heritage Fujian and Taiwan Ministry of Culture and Tourism China Xiamen University Institute of Artificial Intelligence Xiamen361005 China

ISBN: (纸本)9798350386226

Self-supervised pre-training followed by fine-tuning is a potent paradigm for few-shot learning, leveraging extensive unlabeled data with remarkable efficacy. Current self-supervised methods often lean towards Vision Transformers (ViTs) rather than CNN-Transformer hybrid architectures, which generally demonstrate superior performance. However, this reliance on ViTs can lead to poor perception of local features by the model. The challenge lies in designing a suitable proxy task for hybrid architectures like CNN-Transformers, which have significant structural differences. Additionally, the current organization of CNN-Transformer hybrid backbones is often sequential, hindering collaboration during pre-training and the acquisition of robust representations. To address these issues, we propose Self-Supervised Collaborative CNN-Transformer (S2CCT) for few-shot medical image segmentation. This framework introduces three innovative designs: (1) a composite proxy task based on image masking and image super-resolution tailored for CNN-Transformer hybrid architectures, enabling the backbone to acquire robust representations during pre-training that can be transferred to downstream tasks;(2) a parallel CNN-Transformer architecture that better attends to multi-scale features in images, making it more suitable for dense prediction tasks like image segmentation;(3) a sparse and dense feature fusion module to enhance collaboration between the two encoders. Experiments demonstrate that S2CCT outperforms previous state-of-the-art methods on two public medical image segmentation benchmarks, i.e., ACDC and KiTs19. The code and pretrained models will be released soon. © 2024 IEEE.

关键词： Image segmentation

来源：评论

学校读者我要写书评

暂无评论

Effects of spray adjuvants and operation modes on droplet deposition and elm aphid control using an unmanned aerial vehicle

引用

International Journal of Agricultural and Biological Engineering 2023年第2期16卷 1-9页

作者： Zechen Dou Zhihao Fang Xiaoqiang Han Muhammad Zeeshan Yapeng Liu Yubin Lan College of Agriculture/Key Laboratory of Oasis Agricultural Pest Management and Plant Protection Resources Utilization Shihezi UniversityShihezi 832003XinjiangChina College of Agricultural Engineering and Food Science Shandong University of TechnologyZibo 255049ShandongChina Shandong Provincial Engineering Technology Research Center for Agricultural Aviation Intelligent Equipment Zibo 255049ShandongChina College of Electronics Engineering(College of Artificial Intelligence) South China Agricultural UniversityGuangzhou 510642China

A conventional spraying mode and a fully autonomous fruit tree operation mode using a model DJ T30 unmanned aerial vehicle(UAV)were used to control aphids control on elm trees and to clarify the distribution of droplets in elm trees sprayed by a *** effects of six aviation spray adjuvants on elm canopy droplet deposition and aphid control were *** software was used to analyze and measure the droplet density and deposition of water sensitive paper in two modes;this was done to calculate the droplet uniformity,depositional penetration,and droplet penetration,and to verify the aphid control *** results showed that the droplet density increased by 79.7%-100.7% in the upper canopy and 0-394.1%in the lower canopy without adjuvants in the fully autonomous fruit tree operation *** upper canopy deposits increased by 65.7%-179.3%,and the lower canopy increased by 0-152.8%.When adjuvants were added,the droplet density in the upper canopy increased by 49.7-56.1%using Jiexiaofeng(JXF),and the lower canopy increased by 138.2%-177.8% using JXF,45.8%-141.3%using Beidatong(BDT),45.5%-92.9% using Gongbei(GB),0-93.5% using Maisi(MS),and 0-95.2%using Manniu(MN).The deposits of the upper canopy increased by 888.1-1154.2% using JXF,0-1298.3% using MN,0-343.9%using BDT,0-422.5% using GB,0-580.3% using *** lower canopy increased by 746.4%-1426.0%using JXF,226.2%-231.0% using BDT,435.8%-644.0% using GB,255.0%-322.4%using MS,and 249.3%-360.0%using *** JXF was added,the droplet uniformity,droplet penetration and depositional penetration were better than when using other *** effects of JXF,BDT and GB in controlling aphids was significantly better than other adjuvants(p<0.05).The following control effects were observed;94.1% with JXF,93.1% with BDT,and 93.3% with GB after 3 d of application,and 97.9% with JXF,95.6% with BDT,and 97.1% with GB after 7 d of *** the same time,the application of the fully autonomous fruit tree operation mode a

关键词： adjuvants UAV operation mode droplet deposition aphid

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：