检索结果-内蒙古大学图书馆

SSRN 2023年

作者： Wang, Lizhi Lin, Xue Liu, Ximing Gong, Xiaoli Zhang, Jin Wang, Wenwen Yew, Pen-Chung College of Computer Science Nankai University Tianjin300350 China Department of Computer Science University of Georgia AthensGA30602-7404 United States Department of Computer Science and Engineering University of Minnesota at Twin Cities MinneapolisMN55455 United States Tianjin Key Laboratory of Brain Science and Intelligent Rehabilitation Tianjin China

Current hardware technologies, such as Intel SGX and RDMA, rely on swapping architectures to maintain compatibility with existing systems. However, the high bandwidth and low latency of these hardware designs shift the bottleneck of swappingfrom low-speed devices to physical transmission carriers and virtual conversion processes. Furthermore, the unavoidable worldswitch introduces significant overhead during page fault processing, leading to significant performance penalties, especially for memory-intensive applications with large memory footprints. Thus, it is crucial to explore optimization opportunities to improvethe efficiency of swapping architectures. In this paper, we propose two effective page preloading schemes, which we have implemented in a prototype with LLVM. Our experimental results on SPEC CPU2017 demonstrate that, on average, the two proposedmechanisms can achieve performance improvements of 16.9% and 11.7%, respectively, with a maximum improvement of 22.0%and 15.3%. Additionally, our combined approach can achieve an improvement of 9.8% on real-world applications such as SIFTand MSER. © 2023, The Authors. All rights reserved.

关键词： Memory architecture

来源：评论

学校读者我要写书评

暂无评论

ICM-Assistant: Instruction-tuning Multimodal Large Language Models for Rule-based Explainable Image Content Moderation

arXiv

引用

arXiv 2024年

作者： Wu, Mengyang Zhao, Yuzhi Cao, Jialun Xu, Mingjie Jiang, Zhongming Wang, Xuehui Li, Qinbin Hu, Guangneng Qin, Shengchao Fu, Chi-Wing Department of Computer Science and Engineering The Chinese University of Hong Kong Hong Kong Huawei Hong Kong Research Center Hong Kong Department of Computer Science and Engineering The Hong Kong University of Science and Technology Hong Kong Huawei 2012 Laboratories Artificial Intelligence Institute Shanghai Jiao Tong University China School of Computer Science and Technology Huazhong University of Science and Technology China School of Computer Science and Technology Xidian University China Guangzhou Institute of Technology Xidian University China ICTT and ISN Laboratory Xidian University China

Controversial contents largely inundate the Internet, infringing various cultural norms and child protection standards. Traditional Image Content Moderation (ICM) models fall short in producing precise moderation decisions for diverse standards, while recent multimodal large language models (MLLMs), when adopted to general rule-based ICM, often produce classification and explanation results that are inconsistent with human moderators. Aiming at flexible, explainable, and accurate ICM, we design a novel rule-based dataset generation pipeline, decomposing concise human-defined rules and leveraging well-designed multi-stage prompts to enrich short explicit image annotations. Our ICM-Instruct dataset includes detailed moderation explanation and moderation Q-A pairs. Built upon it, we create our ICM-Assistant model in the framework of rule-based ICM, making it readily applicable in real practice. Our ICM-Assistant model demonstrates exceptional performance and flexibility. Specifically, it significantly outperforms existing approaches on various sources, improving both the moderation classification (36.8% on average) and moderation explanation quality (26.6% on average) consistently over existing MLLMs. Caution: Content includes offensive language or images. Code — https://***/zhaoyuzhi/ICM-Assistant Copyright © 2024, The Authors. All rights reserved.

关键词： HTTP

来源：评论

学校读者我要写书评

暂无评论

Contamination-Resilient Anomaly Detection via Adversarial Learning on Partially-Observed Normal and Anomalous Data 41

Contamination-Resilient Anomaly Detection via Adversarial Le...

引用

41st International Conference on Machine Learning, ICML 2024

作者： Lv, Wenxi Su, Qinliang Wan, Hai Xu, Hongteng Xu, Wenchao School of Computer Science and Engineering Sun Yat-sen University Guangzhou China Guangdong Key Laboratory of Big Data Analysis and Processing Guangzhou China Gaoling School of Artifical Intelligence Renmin University of China Beijing China Department of Computing The Hong Kong Polytechnic University Hong Kong

Many existing anomaly detection methods assume the availability of a large-scale normal dataset. But for many applications, limited by resources, removing all anomalous samples from a large unlabeled dataset is unrealistic, resulting in contaminated datasets. To detect anomalies accurately under such scenarios, from the probabilistic perspective, the key question becomes how to learn the normal-data distribution from a contaminated dataset. To this end, we propose to collect two additional small datasets that are comprised of partially-observed normal and anomaly samples, and then use them to help learn the distribution under an adversarial learning scheme. We prove that under some mild conditions, the proposed method is able to learn the correct normal-data distribution. Then, we consider the overfitting issue caused by the small size of the two additional datasets, and a correctness-guaranteed flipping mechanism is further developed to alleviate it. Theoretical results under incomplete observed anomaly types are also presented. Extensive experimental results demonstrate that our method outperforms representative baselines when detecting anomalies under contaminated datasets. Copyright 2024 by the author(s)

关键词： Adversarial machine learning

来源：评论

学校读者我要写书评

暂无评论

SwinTaste: Bimodal Biosignals Taste Sensation Recognition via Swin Transformer

SwinTaste: Bimodal Biosignals Taste Sensation Recognition vi...

引用

International Joint Conference on Neural Networks (IJCNN)

作者： Han Gao Shuo Zhao You Wang Jin Zhang Wei Yao Zhiyuan Luo Guang Li State Key Laboratory of Industrial Control Technology Zhejiang University Hangzhou China School of Computer and Communication Engineering Changsha University of Science and Technology Changsha China Department of Computer Science Royal Holloway University of London Egham U.K.

ISBN: (数字)9798350359312

ISBN: (纸本)9798350359329

Objective assessment of taste sensation is essential for medical diagnosis, food development, and multisensory interaction. Human taste sensation can be characterized through biosignals such as electroencephalography (EEG) and electromyography (EMG). However, taste sensation recognition on multiple-subject datasets remains challenging due to the low signal-to-noise ratio and substantial individual variability of biosignals. To address these problems, we propose SwinTaste for accurate and generalized taste sensation recognition from bimodal biosignals. The Transformer is introduced to extract hierarchical features. A two-stage patch partition module is optimized for the characteristics of biosignals. Moreover, a multi-task learning strategy is adopted to improve the generalization and subject adaptation abilities. The SwinTaste is evaluated on a multiple-subject taste sensation dataset. Comparison experiments and ablation studies demonstrate the superior performance of SwinTaste, indicating the potential for generalized application in biosignal recognition.

关键词： Accuracy Sensitivity Biological system modeling Brain modeling Transformers Multitasking Feature extraction

来源：评论

学校读者我要写书评

暂无评论

VeriCache: Formally Verified Fine-Grained Partitioned Cache for Side-Channel-Secure Enclaves

引用

IEEE Transactions on Dependable and Secure Computing 2025年

作者： Yin, Lingfeng Wang, Haixia Lyu, Yongqiang Hu, Chaojian Wang, Dongsheng Tsinghua University Department of Computer Science and Technology Beijing China Zhongguancun Laboratory Beijing China Tsinghua University Beijing National Research Center for Information Science and Technology Beijing China Huawei Technologies Beijing China

Cache side-channel attacks significantly threaten the security of TEE-enabled secure architectures such as Intel SGX due to the shared cache resources between the attacker and victim processes. Fine-grained partitioned caches are presented to prevent these attacks, but state-of-the-art partitioned cache designs are faced with the issue of performance losses caused by low cache capacity utilization in all-enclave scenarios, such as multi-tenant cloud environments. We propose VeriCache, a fine-grained partitioned cache, which can support exclusive partitions for large-scale enclaves and is verified secure against side-channel attacks by abstracted cache behavior model checking. To fully utilize cache capacity, VeriCache designs a two-level partition strategy based on a tagged cache architecture and employs a flexible moving boundary between enclaves and non-enclaves. It supports flexible scenarios of the partitions from zero enclaves to all enclaves. We implement VeriCache in the Gem5 simulator for performance evaluation. For a single enclave with a 224KB partition, VeriCache incurs a performance overhead of 11.4% and is slightly better than state-of-the-art partitioned cache, Composable Cachelets (12.7%). In all-enclave scenarios, VeriCache can improve the average performance by 2.7% and increase the LLC hit rates by 7.6% compared to Composable Cachelets. © 2004-2012 IEEE.

关键词： Cache memory

来源：评论

学校读者我要写书评

暂无评论

An Enhanced Data-Driven Model for Lithium-Ion Battery State-of-Health Estimation with Optimized Features and Prior Knowledge

引用

Automotive Innovation 2022年第2期5卷 134-145页

作者： Huanyang Huang Jinhao Meng Yuhong Wang Lei Cai Jichang Peng Ji Wu Qian Xiao Tianqi Liu Remus Teodorescu College of Electrical Engineering Sichuan UniversityChengduChina Faculty of Computer Science and Engineering Shanxi Key Laboratory for Network Computing and Security TechnologyXi’an University of TechnologyXi’anChina Smart Grid Research Institute Nanjing Institute of TechnologyNanjingChina Department of Automation University of Science and Technology of ChinaHefeiChina Key Laboratory of Smart Grid of Ministry of Education Tianjin UniversityTianjinChina Department of Energy Technology Aalborg University9220 AalborgDenmark

In the long-term prediction of battery degradation,the data-driven method has great potential with historical data recorded by the battery management *** paper proposes an enhanced data-driven model for Lithium-ion(Li-ion)battery state of health(SOH)estimation with a superior modeling procedure and optimized *** Gaussian process regression(GPR)method is adopted to establish the data-driven estimator,which enables Li-ion battery SOH estimation with the uncertainty level.A novel kernel function,with the prior knowledge of Li-ion battery degradation,is then introduced to improve the mod-eling capability of the *** for the features,a two-stage processing structure is proposed to find a suitable partial charging voltage profile with high *** the first stage,an optimal partial charging voltage is selected by the grid search;while in the second stage,the principal component analysis is conducted to increase both estimation accuracy and computing *** of the proposed method are validated on two datasets from different Li-ion batteries:Compared with other methods,the proposed method can achieve the same accuracy level in the Oxford dataset;while in Maryland dataset,the mean absolute error,the root-mean-squared error,and the maximum error are at least improved by 16.36%,32.43%,and 45.46%,respectively.

关键词： Li-ion battery State of health Gaussian process regression Kernel function Feature optimization

来源：评论

学校读者我要写书评

暂无评论

Accuracy of Determining Damage from Atmosphere Pollution by Industry Under Various Combinations of Events 17

Accuracy of Determining Damage from Atmosphere Pollution by ...

引用

17th International Conference on Management of Large-Scale System Development, MLSD 2024

作者： Fursov, Sergey Kushnikova, Elena Kushikov, Oleg Shnaider, Ilia V.A. Trapeznikov Institute of Control Sciences of Ras Laboratory #33 Moscow Russia Institute for Problems of Precision Mechanics and Control A Separate Structural Unit of the Federal Research Center Ssc Ras Saratov Russia Saratov State University Department of Mathematical Cybernetics and Computer Science Saratov Russia

ISBN: (纸本)9798350375718

The article develops a mathematical model that allows us to consider the influence of critical combinations of events solving the problem of minimizing damage from the impact of air pollutants on the population, agriculture and the environment. The results obtained can be used to improve environmental monitoring systems. © 2024 IEEE.

关键词： Environmental monitoring

来源：评论

学校读者我要写书评

暂无评论

Enhanced Medical Image Authentication and Precise Tamper Localization Using Discrete Cosine Stockwell Transform

Enhanced Medical Image Authentication and Precise Tamper Loc...

引用

Modeling, Simulation & Intelligent Computing (MoSICom), International Conference on

作者： Riadh Bouarroudj Fouzi Harrou Nabil Zerrouki Feryel Souami Fatma Zohra Bellala Ying Sun Department of Computer Science University of Science and Technology Houari Boumedienne Algiers Algeria CEMSE Division King Abdullah University of Science and Technology Thuwal Saudi Arabia DI2M Laboratory Center for Development of Advanced Technology Algiers Algeria Department of Computer Science University of Algiers Benyoucef Benkhedda Algiers Algeria

ISBN: (数字)9798331533311

ISBN: (纸本)9798331533328

Advancements in communication have boosted digital image sharing but also heightened security risks due to easy manipulation. Therefore, authentication techniques are crucial for securing medical images. This paper proposes a fully blind fragile watermarking model for color and grayscale medical image authentication and accurate tamper localization. The watermark is initially derived from the host image using the Discrete Wavelet Transform (DWT) and encrypted by a Logistic map-based technique to enhance the model’s security. The image is then divided into $4 \times 4$ blocks, and the Discrete Cosine Stockwell Transform (DCST) is performed on each block. Afterward, QR decomposition is performed on each frequency subband, and the watermark is inserted within the R matrices. Experimental results across various types of medical images indicate that the proposed technique provides high watermarked image quality with a Peak Signal to Noise Ratio (PSNR) exceeding 51dB while maintaining good embedding capacity and high sensitivity to attacks. Furthermore, the scheme achieves a highly accurate tamper localization, detecting tampering rates as low as $0.01 \%$ of the image size, which surpasses the accuracy achieved by other models in the literature.

关键词： Location awareness Accuracy Sensitivity PSNR Authentication Watermarking Transforms Discrete wavelet transforms Matrix decomposition Biomedical imaging

来源：评论

学校读者我要写书评

暂无评论

PDDepth: Pose Decoupled Monocular Depth Estimation for Roadside Perception System

引用

IEEE Transactions on Circuits and Systems for Video Technology 2025年

作者： Wang, Huanan Zhang, Xinyu Chen, Zhengxian Jun, Li Liu, Huaping School of Vehicle and Mobility Tsinghua University Beijing100084 China State Key Laboratory of Intelligent Technology and Systems Department of Computer Science and Technology Tsinghua University Beijing100084 China

Accurate depth information is crucial for roadside perception in Cooperative Vehicle Infrastructure Systems. Beyond existing radar and LiDAR solutions, monocular depth estimation using surveillance cameras is emerging as a superior approach due to its cost-effectiveness and dense depth output. Unlike onboard cameras, roadside cameras are relatively fixed in position. Many existing monocular depth estimation methods, which do not independently model camera pose, tend to overfit to training data and produce suboptimal results when confronted with slight variations in camera poses, which may be caused by external forces within the same camera or across different cameras. To address this issue, a pose decoupled monocular depth estimation method specifically designed for roadside perception systems is proposed. This method separates depth estimation into two components: a pose-dependent modeling portion that recovers ground depth based on the current camera pose, and a pose-agnostic portion that estimates pixel height relative to the ground plane. Additionally, a knowledge distillation framework is introduced to improve the robustness of the proposed method against variations in roadside cameras. To validate the method, we propose the first open source dataset for roadside monocular depth estimation, DAIR-MDE, and a roadside instance segmentation dataset, DAIR-Ins, both derived from the DAIR dataset. The proposed method demonstrates significant advances over the state-of-the-art methods on DAIR-MDE. The proposed dataset and source code are publicly available at https://***/441599828/PDDepth. ©1991-2012 IEEE.

关键词： Roadsides

来源：评论

学校读者我要写书评

暂无评论

End-to-end Open-vocabulary Video Visual Relationship Detection using Multi-modal Prompting

arXiv

引用

arXiv 2024年

作者： Wang, Yongqi Yang, Shuo Wu, Xinxiao Luo, Jiebo The Beijing Key Laboratory of Intelligent Information Technology School of Computer Science and Technology Beijing Institute of Technology Beijing100081 China The Guangdong Provincial Laboratory of Machine Perception and Intelligent Computing Shenzhen MSU-BIT University Shenzhen518172 China The Beijing Laboratory of Intelligent Information Technology School of Computer Science and Technology Beijing Institute of Technology Beijing100081 China The Department of Computer Science University of Rochester RochesterNY14627 United States

Open-vocabulary video visual relationship detection aims to expand video visual relationship detection beyond annotated categories by detecting unseen relationships between both seen and unseen objects in videos. Existing methods usually use trajectory detectors trained on closed datasets to detect object trajectories, and then feed these trajectories into large-scale pre-trained vision-language models to achieve open-vocabulary classification. Such heavy dependence on the pre-trained trajectory detectors limits their ability to generalize to novel object categories, leading to performance degradation. To address this challenge, we propose to unify object trajectory detection and relationship classification into an end-to-end open-vocabulary framework. Under this framework, we propose a relationship-aware open-vocabulary trajectory detector. It primarily consists of a query-based Transformer decoder, where the visual encoder of CLIP is distilled for frame-wise open-vocabulary object detection, and a trajectory associator. To exploit relationship context during trajectory detection, a relationship query is embedded into the Transformer decoder, and accordingly, an auxiliary relationship loss is designed to enable the decoder to perceive the relationships between objects explicitly. Moreover, we propose an open-vocabulary relationship classifier that leverages the rich semantic knowledge of CLIP to discover novel relationships. To adapt CLIP well to relationship classification, we design a multi-modal prompting method that employs spatio-temporal visual prompting for visual representation and vision-guided language prompting for language input. Extensive experiments on two public datasets, VidVRD and VidOR, demonstrate the effectiveness of our framework. Our framework is also applied to a more difficult cross-dataset scenario to further demonstrate its generalization ability. © 2024, CC BY.

关键词： Decoding

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：