检索结果-内蒙古大学图书馆

Align is not Enough: Multimodal Universal Jailbreak Attack against Multimodal Large Language Models

IEEE Transactions on Circuits and Systems for Video technology 2025年第6期35卷 5475-5488页

作者： Wang, Youze Hu, Wenbo Dong, Yinpeng Liu, Jing Zhang, Hanwang Hong, Richang Hefei University of Technology School of Computer Science and Information Engineering Hefei230009 China Tsinghua University Beijing100084 China Chinese academy of science Institute of automation Beijing100190 China Nanyang Technological University School of Computer Science and Engineering 639798 Singapore

Large Language Models (LLMs) have evolved into Multimodal Large Language Models (MLLMs), significantly enhancing their capabilities by integrating visual information and other types, thus aligning more closely with the nature of human intelligence, which processes a variety of data forms beyond just text. Despite advancements, the undesirable generation of these models remains a critical concern, particularly due to vulnerabilities exposed by text-based jailbreak attacks, which have represented a significant threat by challenging existing safety protocols. Motivated by the unique security risks posed by the integration of new and old modalities for MLLMs, we propose a unified multimodal universal jailbreak attack framework that leverages iterative image-text interactions and transfer-based strategy to generate a universal adversarial suffix and image. Our work not only highlights the interaction of image-text modalities can be used as a critical vulnerability but also validates that multimodal universal jailbreak attacks can bring higher-quality undesirable generations across different MLLMs. We evaluate the undesirable context generation of MLLMs like LLaVA, Yi-VL, MiniGPT4, MiniGPT-v2, and InstructBLIP, and reveal significant multimodal safety alignment issues, highlighting the inadequacy of current safety mechanisms against sophisticated multimodal attacks. This study underscores the urgent need for robust safety measures in MLLMs, advocating for a comprehensive review and enhancement of security protocols to mitigate potential risks associated with multimodal capabilities. © 1991-2012 IEEE.

关键词： Human form models

来源：评论

学校读者我要写书评

暂无评论

JNMR: Joint Non-Linear Motion Regression for Video Frame Interpolation

引用

IEEE Transactions on Image Processing 2023年 32卷 5283-5295页

作者： Liu, Meiqin Xu, Chenming Yao, Chao Lin, Chunyu Zhao, Yao Beijing Jiaotong University Institute of Information Science Beijing100044 China Beijing Key Laboratory of Advanced Information Science and Network Technology Beijing100044 China University of Science and Technology Beijing School of Computer and Communication Engineering Beijing100083 China

Video frame interpolation (VFI) aims to generate predictive frames by motion-warping from bidirectional references. Most examples of VFI utilize spatiotemporal semantic information to realize motion estimation and interpolation. However, due to variable acceleration, irregular movement trajectories, and camera movement in real-world cases, they can not be sufficient to deal with non-linear middle frame estimation. In this paper, we present a reformulation of the VFI as a joint non-linear motion regression (JNMR) strategy to model the complicated inter-frame motions. Specifically, the motion trajectory between the target frame and multiple reference frames is regressed by a temporal concatenation of multi-stage quadratic models. Then, a comprehensive joint distribution is constructed to connect all temporal motions. Moreover, to reserve more contextual details for joint regression, the feature learning network is devised to explore clarified feature expressions with dense skip-connection. Later, a coarse-to-fine synthesis enhancement module is utilized to learn visual dynamics at different resolutions with multi-scale textures. The experimental VFI results show the effectiveness and significant improvement of joint motion regression over the state-of-the-art methods. The code is available at https://***/ruhig6/JNMR. © 1992-2012 IEEE.

关键词： Interpolation

来源：评论

学校读者我要写书评

暂无评论

ET-Net:A Novel Framework for Fine-Grained Traffic Classification in Intelligent Vehicle Applications

引用

China communications 2025年第1期22卷 265-276页

作者： Wei Wenjie Ji Nan Gao Feiran Lin Fuhong School of Computer and Communication Engineering University of Science and Technology BeijingBeijing 100083China China Communications Magazine Co.LTD Beijing 100081China Beijing-Dublin International College Beijing University of TechnologyBeijing 100124China North China Electric Power University Beijing 102206China

Intelligent vehicle applications provide convenience but raise privacy and security *** of sensitive data,including vehicle location,and facial recognition information,poses a threat to user ***,traffic classification is vital for promptly overseeing and controlling applications with sensitive *** this paper,we propose ETNet,a framework that combines multiple features and leverages self-attention mechanisms to learn deep relationships between ***-Net employs a multisimilarity triplet network to extract features from raw bytes,and exploits self-attention to capture long-range dependencies within packets in a session and contextual information ***,we utilizing the loss function to more effectively integrate information acquired from both byte sequences and their corresponding *** simulated evaluations on datasets with similar attributes,ET-Net demonstrates the ability to finely distinguish between nine categories of applications,achieving superior results compared to existing methods.

关键词： attention mechanism encrypted traffic classification intelligent vehicles privacy and security

来源：评论

学校读者我要写书评

暂无评论

Belief Propagation List Decoding for Polar Codes:Performance Analysis and Software Implementation on GPU

引用

China communications 2023年第9期20卷 115-126页

作者： Zhanxian Liu Wei Li Lei Sun Wei Li Jianquan Wang Haijun Zhang School of Computer&Communication Engineering University of Science and Technology BeijingBeijing 100083China Satellite Frequency and Orbit Resource Agency State Radio Monitoring CenterBeijing 100041China School of Automation and Electrical Engineering University of Science and Technology BeijingBeijing 100083China Beijing Engineering and Technology Research Center for Convergence Networks and Ubiquitous Services Institute of Artificial IntelligenceUniversity of Science and Technology BeijingBeijing 100083China

Belief propagation(BP)decoding outputs soft information and can be naturally used in iterative *** list(BPL)decoding provides comparable error-correction performance to the successive cancellation list(SCL)*** this paper,we firstly introduce an enhanced code construction scheme for BPL decoding to improve its errorcorrection ***,a GPU-based BPL decoder with adoption of the new code construction is ***,the proposed BPL decoder is tested on NVIDIA RTX3070 and *** results show that the presented BPL decoder with early termination criterion achieves above 1 Gbps throughput on RTX3070 for the code(1024,512)with 32 lists under good channel conditions.

关键词： polar code belief propagation SIMT list decoding GPU

来源：评论

学校读者我要写书评

暂无评论

FERMixNet: An Occlusion Robust Facial Expression Recognition Model with Facial Mixing Augmentation and Mid-Level Representation Learning

引用

IEEE Transactions on Affective Computing 2024年第2期16卷 639-654页

作者： Huang, Yansong Peng, Junjie Zhang, Wenqiang Zhao, Tong Chen, Gan Tan, Shuhua Yi, Fen Wang, Lu Shanghai University School of Computer Engineering and Science Shanghai China Shanghai University School of Computer Engineering and Science The Shanghai Institute for Advanced Communication and Data Science Shanghai China Fudan University Academy for Engineering and Technology School of Computer Science and Technology Shanghai China YTO Express Co. Ltd. The National Logistics Engineering Laboratory Shanghai China

Facial expressions can provide a better understanding of people's mental status and attitudes towards specific things. However, facial occlusion in real world is an unfavorable phenomenon that greatly affects the performance of facial expression recognition models. Recent works addressing the occlusion problem have primarily relied on attention mechanisms or occlusion discarding methods that focus on non-occluded regions of the face. However, these methods have not achieved a good balance between occlusion robustness and model efficiency. In this paper, we propose a simple and efficient model, called FERMixNet, for occluded facial expression recognition. The model incorporates a novel facial mixing augmentation strategy (FERMix) that generates new training samples by simulating real-world facial occlusion and preserving high expression-related semantic information. By co-training the original and newly generated samples, the model's occlusion robustness is improved without increasing its complexity during inference. Additionally, to further enhance the model's occlusion robustness, we include mid-level representation learning in the network to learn the discriminative non-occluded local features of the samples with low computational cost. Extensive experiments on four public facial occlusion datasets: Occlusion-RAF-DB, Occlusion-FERPlus and FED-RO show that the proposed model achieves state-of-the-art results which demonstrates the good robustness of our method for occluded facial expression recognition. Meanwhile, the proposed model also achieves state-of-the-art results on the in-the-wild facial expression datasets RAF-DB, AffectNet-8, and AffectNet-7. It proves that the proposed model has good application prospects in real world. © 2010-2012 IEEE.

关键词： Face recognition

来源：评论

学校读者我要写书评

暂无评论

Survey of neurocognitive disorder detection methods based on speech, visual, and virtual reality technologies

引用

虚拟现实与智能硬件（中英文） 2024年第6期6卷 421-472页

作者： Tian ZHENG Xinheng WANG Xiaolan PENG Ning SU Tianyi XU Xurong XIE Jin HUANG Lun XIE Feng TIAN Institute of Sofiware Chinese Academy of SciencesBeijing 100190China School of Computer&Communication Engineering University of Science and Technology BeijingBeijing 100083China School of Computer Science and Technology University of Chinese Academy of SciencesBeijing 100049China Peking Union Medical College Hospital Peking Union Medical College and Chinese Academy of Medical ScienceBeijing 100005China

The global trend of population aging poses significant challenges to society and healthcare systems,particularly because of neurocognitive disorders(NCDs)such as Parkinson's disease(PD)and Alzheimer's disease(AD).In this context,artificial intelligence techniques have demonstrated promising potential for the objective assessment and detection of *** contactless screening technologies,such as speech-language processing,computer vision,and virtual reality,offer efficient and convenient methods for disease diagnosis and progression *** paper systematically reviews the specific methods and applications of these technologies in the detection of NCDs using data collection paradigms,feature extraction,and modeling ***,the potential applications and future prospects of these technologies for the detection of cognitive and motor disorders are *** providing a comprehensive summary and refinement of the extant theories,methodologies,and applications,this study aims to facilitate an in-depth understanding of these technologies for researchers,both within and outside the *** the best of our knowledge,this is the first survey to cover the use of speech-language processing,computer vision,and virtual reality technologies for the detection of NSDs.

关键词： Neurocognitive disorder Speech-language computer vision Virtual reality Artificial intelligence

来源：评论

学校读者我要写书评

暂无评论

3D reconstruction of movable cultural relics based on salient region optimization

引用

The Journal of China Universities of Posts and Telecommunications 2023年第5期30卷 11-31页

作者： Wang Wenhao Zhao Haiying School of Information and Communication Engineering Beijing Information Science and Technology UniversityBeijing 100192China School of Artificial Intelligence Beijing University of Posts and TelecommunicationsBeijing 100876China

How to protect cultural retics is of great significance to the transmission and dissemination of history and *** 3-dimensional(3D)modeling of cultural relics is an effective way to preserve *** efficiency and complexity of cultural relic model reconstruction algorithms are significant challenges due to redundant *** tackle the above issue,a 3D reconstruction algorithm,named COLMAP+LSH,was proposed for movable cultural relics based on salient region ***+LSH algorithm introduces saliency region detection and locality-sensetive Hashing(LSH)to achieve efficient,accurate,and robust digital 3D modeling of cultural ***,400 cultural model data were collected through offline and online ***+LSH algorithm detects the salient region interactively and reduces the number of images in the salient region by feature ***,COLMAP+LSH algorithm utilizes LSH to calculate the image selection scores and employs the image selection scores to reduce the redundant *** experiments on the self-constructed cultural relics dataset show that COLMAP+LSH algorithm can efficiently achieve image feature diffusion and ensure the quality of artifact reconstruction while selecting most of the redundant image data.

关键词： digitization of cultural relics 3-dimensional(3D)reconstruction saliency region detection locality-sensetive Hashing(LSH)

来源：评论

学校读者我要写书评

暂无评论

Multi-Objective Optimization for NOMA-Based Mobile Edge Computing Offloading by Maximizing System Utility

引用

China communications 2023年第12期20卷 156-165页

作者： Hong Qin Haitao Du Huahua Wang Li Su Yunfeng Peng Chongqing University of Post and Telecom Chongqing 400065China China Mobile Research Institute Beijing 100000China School of Computer and Communication Engineering University of Science and Technology BeijingBeijing 100083China

Mobile Edge Computing(MEC)is a technology for the fifth-generation(5G)wireless communications to enable User Equipment(UE)to offload tasks to servers deployed at the edge of ***,taking both delay and energy consumption into consideration in the 5G MEC system is usually complex and ***-orthogonal multiple access(NOMA)enable more UEs to offload their computing tasks to MEC servers using the same spectrum resources to enhance the spectrum efficiency for 5G,which makes the problem even more complex in the NOMA-MEC *** this work,a system utility maximization model is present to NOMA-MEC system,and two optimization algorithms based on Newton method and greedy algorithm respectively are proposed to jointly optimize the computing resource allocation,SIC order,transmission time slot allocation,which can easily achieve a better trade-off between the delay and energy *** simulation results prove that the proposed method is effective for NOMA-MEC systems.

关键词： computation offloading mobile edge computing non-orthogonal multiple access resource allocation

来源：评论

学校读者我要写书评

暂无评论

Age estimation from facial images based on Gabor feature fusion and the CIASO-SA algorithm

引用

CAAI Transactions on Intelligence technology 2023年第2期8卷 518-531页

作者： Di Lu Dapeng Wang Kaiyu Zhang Xiangyuan Zeng School of Measurement and Communication Engineering Harbin University of Science and TechnologyHarbinChina Cheriton School of Computer Science University of WaterlooWaterlooOntarioCanada

Aiming at the problem of long time-consuming and low accuracy of existing age estimation approaches,a new age estimation method using Gabor feature fusion,and an improved atomic search algorithm for feature selection is ***,texture features of five scales and eight directions in the face region are extracted by Gabor wavelet *** statistical histogram is introduced to encode and fuse the directional index with the largest feature value on Gabor ***,a new hybrid feature selection algorithm chaotic improved atom search optimisation with simulated annealing(CIASO-SA)is presented,which is based on an improved atomic search algorithm and the simulated annealing ***,the CIASO-SA algorithm introduces a chaos mechanism during atomic initialisation,significantly improving the convergence speed and accuracy of the ***,a support vector machine(SVM)is used to get classification results of the age *** verify the performance of the proposed algorithm,face images with three resolutions in the Adience dataset are *** the Gabor real part fusion feature at 48�48 resolution,the average accuracy and 1-off accuracy of age classification exhibit a maximum of 60.4%and 85.9%,*** results prove the superiority of the proposed algorithm over the state-of-the-art methods,which is of great referential value for application to the mobile terminals.

关键词： age estimation atom search algorithm feature selection Gabor feature simulated annealing

来源：评论

学校读者我要写书评

暂无评论

ACbot:an IIoT platform for industrial robots

引用

Frontiers of computer science 2025年第4期19卷 13-27页

作者： Rui WANG Xudong MOU Tianyu WO Mingyang ZHANG Yuxin LIU Tiejun WANG Pin LIU Jihong YAN Xudong LIU School of Computer Science and Engineering Beihang UniversityBeijing 100191China Zhongguancun Laboratory Beijing 100194China School of Software Beihang UniversityBeijing 100191China School of Mechatronics Engineering Harbin Institute of TechnologyHarbin 150001China Hangzhou Innovation Institute Beihang UniversityHangzhou 310051China School of Information Engineering China University of GeosciencesBeijing 100083China

As the application of Industrial Robots(IRs)scales and related participants increase,the demands for intelligent Operation and Maintenance(O&M)and multi-tenant collaboration *** methods could no longer cover the requirements,while the Industrial Internet of Things(IIoT)has been considered a promising ***,there’s a lack of IIoT platforms dedicated to IR O&M,including IR maintenance,process optimization,and knowledge *** this context,this paper puts forward the multi-tenant-oriented ACbot platform,which attempts to provide the first holistic IIoT-based solution for O&M of *** on an information model designed for the IR field,ACbot has implemented an application architecture with resource and microservice management across the cloud and multiple *** this basis,we develop four vital applications including real-time monitoring,health management,process optimization,and knowledge *** have deployed the ACbot platform in real-world scenarios that contain various participants,types of IRs,and *** date,ACbot has been accessed by 10 organizations and managed 60 industrial robots,demonstrating that the platform fulfills our ***,the application results also showcase its robustness,versatility,and adaptability for developing and hosting intelligent robot applications.

关键词： IIoT platform industrial robots cloud-edge collaboration intelligent applications

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：