检索结果-内蒙古大学图书馆

arXiv 2025年

作者： Gilkarov, Daniel Dubin, Ran The Department of Computer Science Ariel Cyber Innovation Center Ariel University Ariel Israel The Department of Computer and Software Engineering Ariel Cyber Innovation Center Ariel University Ariel Israel

This paper examines the challenges in distributing AI models through model zoos and file transfer mechanisms. Despite advancements in security measures, vulnerabilities persist, necessitating a multi-layered approach to mitigate risks effectively. The physical security of model files is critical, requiring stringent access controls and attack prevention solutions. This paper proposes a novel solution architecture composed of two prevention approaches. The first is Content Disarm and Reconstruction (CDR), which focuses on disarming serialization attacks that enable attackers to run malicious code as soon as the model is loaded. The second is protecting the model architecture and weights from attacks by using Moving Target Defense (MTD), alerting the model structure, and providing verification steps to detect such attacks. The paper focuses on the highly exploitable Pickle and PyTorch file formats. It demonstrates a 100% disarm rate while validated against known AI model repositories and actual malware attacks from the HuggingFace model zoo. © 2025, CC BY-NC-ND.

关键词： Access control

来源：评论

学校读者我要写书评

暂无评论

Runtime Backdoor Detection for Federated Learning via Representational Dissimilarity Analysis

引用

IEEE Transactions on Dependable and Secure Computing 2025年

作者： Zhang, Xiyue Xue, Xiaoyong Du, Xiaoning Xie, Xiaofei Liu, Yang Sun, Meng University of Oxford Department of Computer Science United Kingdom Peking University Department of Information and Computing Science School of Mathematical Science China Monash University Department of Software Systems and Cybersecurity Australia Singapore Management University School of Computing and Information Systems Singapore Nanyang Technological University School of Computer Science and Engineering Singapore

Federated learning (FL), as a powerful learning paradigm, trains a shared model by aggregating model updates from distributed clients. However, the decoupling of model learning from local data makes FL highly vulnerable to backdoor attacks, where a single compromised client can poison the shared model. While recent progress has been made in backdoor detection, existing methods face challenges with detection accuracy and runtime effectiveness, particularly when dealing with complex model architectures. In this work, we propose a novel approach to detecting malicious clients in an accurate, stable, and efficient manner. Our method utilizes a sampling-based network representation method to quantify dissimilarities between clients, identifying model deviations caused by backdoor injections. We also propose an iterative algorithm to progressively detect and exclude malicious clients as outliers based on these dissimilarity measurements. Evaluations across a range of benchmark tasks demonstrate that our approach outperforms state-of-the-art methods in detection accuracy and defense effectiveness. When deployed for runtime protection, our approach effectively eliminates backdoor injections with marginal overheads. © 2025 IEEE.

关键词： Federated learning

来源：评论

学校读者我要写书评

暂无评论

ATM-NeRF: Accelerating Training for NeRF Rendering on Mobile Devices via Geometric Regularization

引用

IEEE Transactions on Multimedia 2025年

作者： Chen, Yang Zhang, Lin Zhao, Shengjie Zhou, Yicong Tongji University School of Software Engineering Shanghai201804 China Ministry of Education Engineering Research Center of Key Software Technologies for Smart City Perception and Planning Shanghai201804 China University of Macau Department of Computer and Information Science 999078 China

Recently, an increasing number of researchers have been dedicated to transferring the impressive novel view synthesis capability of Neural Radiance Fields (NeRF) to resource-constrained mobile devices. One common solution is to pre-train NeRF and bake it into textured meshes which are well supported by mobile graphics hardware. However, the training process of existing methods often requires several hours even with multiple high-end NVIDIA V100 GPUs. The underlying reason is that these schemes mainly rely on photometric rendering loss, neglecting the geometric relationship between the pre-trained NeRF and the baked results. Standing on this point, we present ATM-NeRF (Accelerating Training for Mobile rendering based on NeRF), which is the first to apply effective geometric regularization constraints during both the pre-training and the baking training stages for faster convergence. Specifically, in the initial NeRF pre-training stage, we enforce consistency of the multi-resolution density grids representing the scene geometry to mitigate the shape-radiance ambiguity problem to some extent, achieving a coarse mesh with smoothness. In the second stage, we utilize the positions and geometric features of 3D points projected from the pre-trained posed depths to provide geometric supervision for joint refinement of geometry and appearance of the coarse mesh. As a result, our ATM-NeRF achieves comparable rendering quality to MobileNeRF with a training speed that is about 30× ∼ 70× faster while maintaining finer structure details of the exported mesh. Our source code and the demo video have been released at https://***/ATM-NeRF/. © 1999-2012 IEEE.

关键词： Mesh generation

来源：评论

学校读者我要写书评

暂无评论

MCDS: An Effective Multi-UAV Collaborative Decision-Making System in Mobile Edge Computing Networks

引用

IEEE Internet of Things Journal 2025年

作者： Xiong, Naixue Zhong, Weiyu Chen, Yuxiang He, Dacheng Li, Yuhui Chen, Linshu Liang, Wei Sanya Research Institute of Hunan University of Science and Technology School of Computer Science and Engineering Hunan University of Science and Technology Hunan Key Laboratory for Service Computing and Novel Software Technology China Sanya Research Institute of Hunan University of Science and Technology Hunan University School of Computer Science and Engineering Hunan University of Science and Technology Hunan Key Laboratory for Service Computing and Novel Software Technology School of Information Science and Engineering China The Hong Kong Polytechnic University Department of Computing Hong Kong

Mobile Edge Computing (MEC) pushes computing resources from the network center to the network edge to provide efficient and reliable computing services. However, due to the mobility and diversity of mobile users (MUs), edge servers (ESs) are likely to be overloaded, which leads to a rapid decline in the quality of service provided by the MEC system. This paper introduces unmanned aerial vehicles (UAVs) to solve the problem of ES overload. This paper presents an effective Multi-UAV Collaborative Decision-Making System (MCDS) tailored to optimize task offloading and resource allocation in MEC networks. The proposed system seeks to reduce task computing delays and energy consumption while guaranteeing timely task completion and compliance with resource constraints. We design a distributed two-stage optimization algorithm to jointly optimize the task offloading decision and resource allocation of the collaborative computing system (UAVs, ESs, and MUs). To address the flight cooperation problem of multi-UAVs, we also proposed a task-aware scheduling algorithm for multi-UAVs. In addition, we conducted a large number of simulation experiments. Experimental results show that our two-stage optimization scheme is lower than other benchmark algorithms in terms of task completion delay, energy consumption and total cost of objective function. Specifically, compared with seven baseline algorithms under three different system setting scenarios, our algorithm reduces the total cost by an average of 38.79%, the energy consumption by an average of 40.70%, and the delay by an average of 15.67%. © 2014 IEEE.

关键词： Drones

来源：评论

学校读者我要写书评

暂无评论

Rethinking Polyp Segmentation from the Perspectives of Matching Views and Seeking Camouflage

引用

Multimedia Systems 2025年第3期31卷 1-17页

作者： Jiang, Zhengfang Chen, Haipeng Yang, Yongping Liu, Xianzhu Lyu, Yingda College of Software Jilin University Changchun130012 China Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education Jilin University Changchun130012 China College of Computer Science and Technology Jilin University Changchun130012 China Department of Colorectal Anal Surgery The Second Hospital of Jilin University Changchun130022 China National and Local Joint Engineering Research Center of Space Optoelectronics Technology Changchun University of Science and Technology Changchun130022 China College of Opto-Electronic Engineering Changchun University of Science and Technology Changchun130022 China Public Computer Education and Research Center Jilin University Changchun130012 China

Accurate segmentation of polyps from colonoscopy images is of great significance for the prevention of colorectal cancer. Despite notable progress, polyp segmentation remains a challenging task due to the diversity of polyp size and shape, as well as the camouflage property of polyps caused by their high similarity to surrounding tissues. To address these challenges, inspired by human perceptual experience in finding objects, we propose a multi-perspective collaborative network (MPCNet), innovatively identifying the potential polyps from the perspectives of matching views and seeking camouflage. Firstly, from the former perspective, a global–local aware module is proposed to provide various feature receptive fields to match suitable views for polyps of different sizes and shapes. Then, from the latter perspective, a texture enhancement module (TEM) and an interaction refinement module (IRM) are proposed to mine and complement clues essential for breaking the camouflage. Specifically, TEM mines texture-related clues and amplifies the subtle texture differences between polyps and their background, while IRM complements boundary clues and further optimizes feature representation through the interaction between boundaries and regions, thereby generating accurate segmentation results with complete regions and fine boundaries. Extensive experiments on five benchmark polyp datasets show that our MPCNet outperforms other state-of-the-art methods. In particular, we achieve a mean Dice of 81.7% and 82.4% on the most challenging datasets CVC-ColonDB and ETIS, respectively, which are significantly higher than competing methods. © The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature 2025.

关键词： Image segmentation

来源：评论

学校读者我要写书评

暂无评论

Predicting Life Expectancy Using Stacking Regressor Models: Analyzing the Impact of GDP and Population Dynamics

Predicting Life Expectancy Using Stacking Regressor Models: ...

引用

International Conference on Electrical, computer and Communication engineering (ECCE)

作者： Marzia Ahmed Abu Kowshir Bitto Md Maruf Hassan Touhid Bhuiyan Department of Software Engineering Daffodil International University Dhaka Bangladesh Department of Computer Science and Engineering Southeast University Dhaka Bangladesh School of Information Technology Washington University of Science and Technology Virginia United States

ISBN: (数字)9798350357509

ISBN: (纸本)9798350357516

Accurately modeling the relationship between socio-economic factors such as GDP, population size, and life expectancy is vital for evidence-based policymaking. This study presents a machine learning framework using stacking regressors to predict life expectancy while capturing the complex, non-linear interactions of GDP and population dynamics over time. The methodology involves comprehensive data preprocessing, parameter tuning via GridSearchCV, and rigorous evaluation using cross-validation to ensure reliability. The proposed approach provides interpretable insights into how variations in GDP and population size influence life expectancy trends, offering valuable guidance for economic and demographic planning. Experimental results highlight the effectiveness of the stacking model, achieving a Mean Squared Error (MSE) of 0.295 and a Mean Absolute Error (MAE) of 0.443, outperforming traditional regression and tree-based methods. This framework offers a robust solution for researchers and policymakers aiming to understand and predict long-term socio-economic outcomes.

关键词： Accuracy Economic indicators Stacking Predictive models Market research Reliability Socioeconomics Random forests Tuning Regression tree analysis

来源：评论

学校读者我要写书评

暂无评论

GLoT: A Novel Gated-Logarithmic Transformer for Efficient Sign Language Translation

arXiv

引用

arXiv 2025年

作者： Shahin, Nada Ismail, Leila Lab Department of Computer Science and Software Engineering College of Information Technology UAE University Al-Ain United Arab Emirates Lab Department of Computer Science and Software Engineering College of Information Technology United Arab Emirates Emirates Center for Mobility Research UAE University Al-Ain United Arab Emirates

Machine Translation has played a critical role in reducing language barriers, but its adaptation for Sign Language Machine Translation (SLMT) has been less explored. Existing works on SLMT mostly use the Transformer neural network which exhibits low performance due to the dynamic nature of the sign language. In this paper, we propose a novel Gated-Logarithmic Transformer (GLoT) that captures the long-term temporal dependencies of the sign language as a time-series data. We perform a comprehensive evaluation of GloT with the transformer and transformer-fusion models as a baseline, for Sign-to-Gloss-to-Text translation. Our results demonstrate that GLoT consistently outperforms the other models across all metrics. These findings underscore its potential to address the communication challenges faced by the Deaf and Hard of Hearing community. © 2025, CC BY.

关键词： Neural machine translation

来源：评论

学校读者我要写书评

暂无评论

TS-RePSO: A Three-Stage Feature Selection Method Combing ReliefF and PSO in Bioinformatics

IEEE Transactions on Computational Biology and Bioinformatic...

引用

IEEE Transactions on Computational Biology and Bioinformatics 2025年第2期22卷 640-652页

作者： Bin Pu Haining Wang Zhaozhao Xu Fangyuan Yang Xiangqiong Wu Jianguo Chen College of Computer Science and Electronic Engineering Hunan University Changsha China School of Software South China University of Technology GuangZhou China School of Computer Science and Technology Henan Polytechnic University Jiaozuo China Department of Gynecologic Oncology The First Affiliated Hospital of Henan Polytechnic University Jiaozuo China School of Software Engineering Sun Yat-Sen University Guangzhou China

The inherent characteristics of high-dimensional feature redundancy of biomedical data lead to the “curse of dimensionality” in bioinformatics, which brings new challenges to feature selection problems. Recently, the two-stage approach combining the filter and wrapper methods has become popular for feature selection tasks. However, these two-stage or previous one-stage algorithms suffer from blindness in the setting of thresholds, and the search methods tend to fall into local optimum solutions. To this end, we propose a three-stage feature selection method that combines ReliefF and Particle swarm optimization as a specific case, called TS-RePSO, including the filter stage, grouping stage, and wrapper stage. Specifically, in the filter stage, ReliefF is utilized to compute the weights of the features and sort them in descending order. In the grouping stage, the ranked features are grouped based on the density equalization strategy so that the weight of groups in all groups is equal. In the wrapper stage, the proposed grouping PSO is employed to search for the grouped features and select them according to the in-group and out-group evaluation strategies. Extensive experiments are conducted on 5 benchmark datasets and 6 real-world datasets, and experiment results show that the proposed method achieves the best performance.

关键词： Filtering algorithms Classification algorithms Bioinformatics Particle swarm optimization Feature extraction Diseases Prediction algorithms Information filters Genetic algorithms Diversity reception

来源：评论

学校读者我要写书评

暂无评论

VisioSLR: A Vision Data-Driven Framework for Sign Language Video Recognition and Performance Evaluation on Fine-Tuned YOLO Models

引用

Procedia computer science 2025年 257卷 85-92页

作者： Leila Ismail Nada Shahin Henok Tesfaye Alain Hennebelle Intelligent Distributed Computing and Systems (INDUCE) Lab Department of Computer Science and Software Engineering College of Information Technology United Arab Emirates University United Arab Emirates Independent Researcher

With the emergence of AI for good, there has been an increasing interest in building computer vision data-driven deep learning inclusive AI solutions. Sign language Recognition (SLR) has gained attention recently. It is an essential component of a sign-to-text translation system to support the deaf and hard-of-hearing population. This paper presents a computer VISIOn data-driven deep learning framework for Sign Language video Recognition (VisoSLR). VisioSLR provides a precise measurement of translating signs for developing an end-to-end computational translation system. Considering the scarcity of sign language datasets, which hinders the development of an accurate recognition model, we evaluate the performance of our framework by fine-tuning the very well-known YOLO models, which are built from a signs-unrelated collection of images and videos, using a small-sized sign language dataset. Gathering a sign language dataset for signs training would involve an enormous amount of time to collect and annotate videos in different environmental setups and multiple signers, in addition to the training time of a model. Numerical evaluations of VisioSLR show that our framework recognizes signs with a mean average precision of 97.4%, 97.1%, and 95.5% and 11, 12, and 12 milliseconds of recognition time on YOLOv8m, YOLOv9m, and YOLOv11m, respectively.

关键词： Artificial intelligence computer vision Deep learning Fine-tuning Machine translation Neural machine translation Sign language Sign language recognition Transfer learning YOLO

来源：评论

学校读者我要写书评

暂无评论

PIP: Perturbation-based Iterative Pruning for Large Language Models

arXiv

引用

arXiv 2025年

作者： Cao, Yi Xu, Wei-Jie Shen, Yucheng Shi, Weijie Chan, Chi-Min Xu, Jiajie School of Computer Science and Technology Soochow University China State Key Laboratory for Novel Software Technology Nanjing University China Department of Computer Science and Engineering The Hong Kong University of Science and Technology Hong Kong

The rapid increase in the parameter counts of Large Language Models (LLMs), reaching billions or even trillions, presents significant challenges for their practical deployment, particularly in resource-constrained environments. To ease this issue, we propose PIP (Perturbation-based Iterative Pruning), a novel double-view structured pruning method to optimize LLMs, which combines information from two different views: the unperturbed view and the perturbed view. With the calculation of gradient differences, PIP iteratively prunes those that struggle to distinguish between these two views. Our experiments show that PIP reduces the parameter count by approximately 20% while retaining over 85% of the original model’s accuracy across varied benchmarks. In some cases, the performance of the pruned model is within 5% of the unpruned version, demonstrating PIP’s ability to preserve key aspects of model effectiveness. Moreover, PIP consistently outperforms existing state-of-the-art (SOTA) structured pruning methods, establishing it as a leading technique for optimizing LLMs in environments with constrained resources. Our code is available at: https://***/caoyiiiiii/PIP. Copyright © 2025, The Authors. All rights reserved.

关键词：

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：