检索结果-内蒙古大学图书馆

IEEE Transactions on Artificial Intelligence 2025年第2期6卷 405-415页

作者： Cai, Zinuo Ma, Rongbo Fu, Yicheng Zhang, Weishan Ma, Ruhui Guan, Haibing Shanghai Jiao Tong University School of Electronic Information and Electrical Engineering Shanghai200240 China Qingdao Institute of Software College of Computer Science and Technology Qingdao266555 China

In recent years, the emergence of large-language models (LLMs) has profoundly transformed our production and lifestyle. These models have shown tremendous potential in fields, such as natural language processing, speech recognition, and recommendation systems, and are increasingly playing crucial roles in applications such as human-computer interaction and intelligent customer service. Efficient inference solutions for LLMs in data centers have been extensively researched, with a focus on meeting users' quality of service requirements. In this article, we focus on two additional requirements that responsible LLM inference should meet under QoS conditions: security throughout the model execution process and low maintenance requirements for the inference system. Therefore, we propose LLMaaS, a trusted model inference platform based on a serverless computing platform aimed at providing inference as a service for LLMs. First, we design a trusted serverless computing platform based on software guard extension (SGX), which includes distributed identity verification and SGX device plugins to ensure the security and trustworthiness of the inference process. Additionally, to reduce the maintenance requirements of the system, we enhance the SGX-based deep learning computing framework, including replacing PyTorch and using a greedy algorithm for graph partitioning. We conduct tests on four typical large models, and the experimental results demonstrate that, with minimal overhead and user code modifications, we can ensure the security of model execution. © 2020 IEEE.

关键词： Quality of service

来源：评论

学校读者我要写书评

暂无评论

OCRBench: on the hidden mystery of OCR in large multimodal models

引用

science China(information sciences) 2024年第12期67卷 23-35页

作者： Yuliang LIU Zhang LI Mingxin HUANG Biao YANG Wenwen YU Chunyuan LI Xu-Cheng YIN Cheng-Lin LIU Lianwen JIN Xiang BAI School of Artificial Intelligence and Automation Huazhong University of Science and Technology School of Electronic and Information Engineering South China University of Technology Microsoft Research School of Computer & Communication Engineering University of Science and Technology Beijing Institute of Automation Chinese Academy of Sciences School of Software Engineering Huazhong University of Science and Technology

Large models have recently played a dominant role in natural language processing and multimodal vision-language learning. However, their effectiveness in text-related visual tasks remains relatively unexplored. In this paper, we conducted a comprehensive evaluation of large multimodal models, such as GPT4V and Gemini, in various text-related visual tasks including text recognition, scene text-centric visual question answering(VQA), document-oriented VQA, key information extraction(KIE), and handwritten mathematical expression recognition(HMER). To facilitate the assessment of optical character recognition(OCR) capabilities in large multimodal models, we propose OCRBench, a comprehensive evaluation benchmark. OCRBench contains 29 datasets, making it the most comprehensive OCR evaluation benchmark available. Furthermore, our study reveals both the strengths and weaknesses of these models, particularly in handling multilingual text, handwritten text, non-semantic text, and mathematical expression *** importantly, the baseline results presented in this study could provide a foundational framework for the conception and assessment of innovative strategies targeted at enhancing zero-shot multimodal *** evaluation pipeline and benchmark are available at https://***/Yuliang-Liu/Multimodal OCR.

关键词： large multimodal model OCR text recognition scene text-centric VQA document-oriented VQA key information extraction handwritten mathematical expression recognition

来源：评论

学校读者我要写书评

暂无评论

Cutting-Edge AI-Driven Higher Educational and Career Advisory Platform

Cutting-Edge AI-Driven Higher Educational and Career Advisor...

引用

2024 International Conference on Innovative Computing, Intelligent Communication and Smart Electrical Systems, ICSES 2024

作者： Dewasurendra, S.V. Udunuwara, U.K.C. Wanniarachchi, T.T. Panditharathne, P.D.R.L. Thelijjagoda, Samantha Dissanayaka, Kapila Sri Lanka Institute of Information Technology Department of Computer Science and Software Engineering Malabe Sri Lanka

ISBN: (纸本)9798331543617

Sri Lankan students, especially those from rural and low-income backgrounds, face significant challenges in accessing higher education, including university applications, career planning, and financial aid. This research introduces an AI-powered platform designed to provide personalized assistance through university course recommendations, career guidance, financial aid information, and AI-generated mock aptitude tests. The platform employs Graph Neural Networks (GNNs) for course recommendations, Recurrent Neural Networks (RNNs) for career guidance, and a Retrieval-Augmented Generation (RAG) model for financial aid navigation. Additionally, AI-driven algorithms with Optical Character Recognition (OCR) technology support exam preparation. Data from students and universities informed model training, and evaluations showed high accuracy and user satisfaction across components. Results indicated that the platform effectively aligns educational resources with students' academic and career goals, promoting greater educational accessibility and support for underserved Sri Lankan students. © 2024 IEEE.

关键词： Graph neural networks

来源：评论

学校读者我要写书评

暂无评论

An attention-based channel estimation network with residual mechanism for 5G OFDM systems 4

An attention-based channel estimation network with residual ...

引用

4th International Conference on Electronic Communication, computer science and Technology, ECCST 2024

作者： Jiang, Jinjing Huang, Weipeng Zhang, Xiao Yang, Kun Peng, Tong School of Information Engineering Zhejiang Ocean University China School of Computer Science and Software Engineering Shenzhen Institute of Information Technology China

Due to the significant increase in data transmission speed and gradual increase in Doppler frequency shift, channel estimation accuracy has become one of the most prioritized considerations in many cases. Specifically, we propose an attention-based channel estimation network with a residual mechanism, named TRCE-Net. In our design, the use of the multi-head attention mechanism at the input of the encoder is to obtain richer feature details. A linear series decoder module is implemented by leveraging multiple residual networks with skip connections. This novel design is capable of improving the channel estimation accuracy after deepening the network while effectively preventing degradation. To lessen the number of parameters that the network needs to learn, we introduce the bi-linear interpolation mechanism to replace the traditional up-sampling layer to mitigate both computational overhead and model complexity. According to the simulation results, our proposed TRCE-Net is superior to other state-of-the-art methods with regard to estimation accuracy and robustness to different maximum Doppler shifts, such as ResNet and InterpolateNet. © Published under licence by IOP Publishing Ltd.

关键词： Orthogonal frequency division multiplexing

来源：评论

学校读者我要写书评

暂无评论

Unified Approach Integrating Machine Learning Algorithms for Detection of e-Mail Phishing Attacks 1

Unified Approach Integrating Machine Learning Algorithms for...

引用

1st International Conference on Computing, sciences and Communications, ICCSC 2024

作者： Mathur, Abeer Dubey, Abhinav Sudhakar Mahara, Pranav Singh Gandhi, Usha Devi School of Computer Science & Engineering Vellore Institute of Technology Department of Software Systems Vellore India School of Computer Science Engineering & Information Systems Vellore Institute of Technology Department of Information Technology Vellore India

ISBN: (纸本)9798350353648

With the ever-rising risk of phishing attacks, which capitalize on vulnerable human behavior in the contemporary digital space, requires new cybersecurity methods. This literary work contributes to the solution by novel incorporation of three techniques: Support Vector Machine (SVM), Natural Language Processing (NLP) and Random Forest (RF). Utilizing NLP for deep analysis of text in phishing e-mails, SVM for strong categorization and Random Forest's decision making by group helps in improving detection. This combination of language and pattern analysis along with group learning builds a fortified system. This approach differs from prior work on a variety of factors such as reliability, security, integrity, scalability, interpretability, transparency, and robustness. Thorough evaluation on a reliable dataset aimed to demonstrate the method's effectiveness, which resulted in 98.7 % of accuracy in detection. This majorly aims at showcasing the potential of ensemble techniques to combat against cyber-attacks like phishing for laying a strong foundation for a secure digital environment. © 2024 IEEE.

关键词： Phishing

来源：评论

学校读者我要写书评

暂无评论

ViF-SD2E: a robust weakly-supervised framework for neural decoding

引用

Neural Computing and Applications 2025年第9期37卷 6645-6661页

作者： Feng, Jingyi Luo, Yong Song, Shuang Hu, Han National Engineering Research Center for Multimedia Software School of Computer Science Wuhan University Wuhan430072 China School of Information and Electronics Beijing Institute of Technology Beijing100081 China

Neural decoding plays a vital role in the interaction between the brain and the outside world. Our task in this paper is to decode the movement track of a finger directly based on the neural data. Existing neural decoding solutions primarily perform some preprocessing operations on neural data before feeding them into existing models (such as LSTM) for decoding. However, these solutions either are prone to overfitting or cannot well exploit the spatial and temporal information. In our previous observations, there is a symmetrical phenomenon between the unsupervised decoded trajectory and the ground truth trajectory within the activity space. This precisely motivates us to propose (or derive) a robust weakly-supervised framework (or model structure), called ViF-SD2E, for neural decoding. In particular, it consists of a space-division (SD) module and an exploration–exploitation (2E) strategy, to effectively exploit both the spatial information of the outside world and the temporal information of neural activity, where the SD2E output is analogized with the weak 0/1 vision feedback (ViF) label for training. Extensive experiments demonstrate the effectiveness of our method, which can sometimes be comparable to supervised counterparts. Therefore, we redirect our attention to the information (hidden in data) ViF-SD2E conveys to us. In other words, we believe that the advantage of ViF-SD2E lies in the fact that its processing steps are objectively determined by the inherent attributes (i.e., symmetry) of the neural data, or rather, the model structure is fixed. © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2025.

关键词： Space division multiple access

来源：评论

学校读者我要写书评

暂无评论

Loop Subgraph-Level Greedy Mapping Algorithm for Grid Coarse-Grained Reconfigurable Array

引用

Tsinghua science and Technology 2023年第2期28卷 330-343页

作者： Naijin Chen Fei Cheng Chenghao Han Jianhui Jiang Xiaoqing Wen School of Computer and Information Science Anhui Polytechnic UniversityWuhu 241000China School of Software Engineering Tongji UniversityShanghai 201804China Department of Computer Science and Networks Kyushu Institute of TechnologyFukuoka 820-8502Japan

To solve the problem of grid coarse-grained reconfigurable array task mapping under multiple constraints,we propose a Loop Subgraph-Level Greedy Mapping(LSLGM)algorithm using parallelism and processing element *** the constraint of a reconfigurable array,the LSLGM algorithm schedules node from a ready queue to the current reconfigurable cell array *** mapping a node,its successor’s indegree value will be dynamically *** its successor’s indegree is zero,it will be directly scheduled to the ready queue;otherwise,the predecessor must be dynamically *** the predecessor cannot be mapped,it will be scheduled to a blocking *** dynamically adjust the ready node scheduling order,the scheduling function is constructed by exploiting factors,such as node number,node level,and node *** with the loop subgraph-level mapping algorithm,experimental results show that the total cycles of the LSLGM algorithm decreases by an average of 33.0%(PEA44)and 33.9%(PEA_(7×7)).Compared with the epimorphism map algorithm,the total cycles of the LSLGM algorithm decrease by an average of 38.1%(PEA_(4×4))and 39.0%(PEA_(7×7)).The feasibility of LSLGM is verified.

关键词： Grid Coarse-Grained Reconfigurable Array(GCGRA) mapping loop subgraph scheduling

来源：评论

学校读者我要写书评

暂无评论

Multi-Objective Equilibrium Optimizer for Feature Selection in High-Dimensional English Speech Emotion Recognition

引用

computers, Materials & Continua 2024年第2期78卷 1957-1975页

作者： Liya Yue Pei Hu Shu-Chuan Chu Jeng-Shyang Pan Fanli Business School Nanyang Institute of TechnologyNanyang473004China School of Computer and Software Nanyang Institute of TechnologyNanyang473004China College of Computer Science and Engineering Shandong University of Science and TechnologyQingdao266590China Department of Information Management Chaoyang University of TechnologyTaichung413310Taiwan

Speech emotion recognition(SER)uses acoustic analysis to find features for emotion recognition and examines variations in voice that are caused by *** number of features acquired with acoustic analysis is extremely high,so we introduce a hybrid filter-wrapper feature selection algorithm based on an improved equilibrium optimizer for constructing an emotion recognition *** proposed algorithm implements multi-objective emotion recognition with the minimum number of selected features and maximum ***,we use the information gain and Fisher Score to sort the features extracted from ***,we employ a multi-objective ranking method to evaluate these features and assign different importance to *** with high rankings have a large probability of being ***,we propose a repair strategy to address the problem of duplicate solutions in multi-objective feature selection,which can improve the diversity of solutions and avoid falling into local *** random forest and K-nearest neighbor classifiers,four English speech emotion datasets are employed to test the proposed algorithm(MBEO)as well as other multi-objective emotion identification *** results illustrate that it performs well in inverted generational distance,hypervolume,Pareto solutions,and execution time,and MBEO is appropriate for high-dimensional English SER.

关键词： Speech emotion recognition filter-wrapper high-dimensional feature selection equilibrium optimizer multi-objective

来源：评论

学校读者我要写书评

暂无评论

Pre-trained SAM as data augmentation for image segmentation

引用

CAAI Transactions on Intelligence Technology 2025年第1期10卷 268-282页

作者： Junjun Wu Yunbo Rao Shaoning Zeng Bob Zhang Yangtze Delta Region Institute(Huzhou) University of Electronic Science and Technology of ChinaHuzhouChina School of Information and Software Engineering University of Electronic Science and Technology of ChinaChengduChina Pattern Analysis and Machine Intelligence Research Group Department of Computer and Information ScienceUniversity of MacaoMacaoChina

Data augmentation plays an important role in training deep neural model by expanding the size and diversity of the ***,data augmentation mainly involved some simple transformations of ***,in order to increase the diversity and complexity of data,more advanced methods appeared and evolved to sophisticated generative ***,these methods required a mass of computation of training or *** this paper,a novel training-free method that utilises the Pre-Trained Segment Anything Model(SAM)model as a data augmentation tool(PTSAM-DA)is proposed to generate the augmented annotations for *** the need for training,it obtains prompt boxes from the original annotations and then feeds the boxes to the pre-trained SAM to generate diverse and improved *** this way,annotations are augmented more ingenious than simple manipulations without incurring huge computation for training a data augmentation *** comparative experiments on three datasets are conducted,including an in-house dataset,ADE20K and *** this in-house dataset,namely Agricultural Plot Segmentation Dataset,maximum improvements of 3.77%and 8.92%are gained in two mainstream metrics,mIoU and mAcc,***,large vision models like SAM are proven to be promising not only in image segmentation but also in data augmentation.

关键词： data augmentation image segmentation large model segment anything model

来源：评论

学校读者我要写书评

暂无评论

DBKEM-AACS: a distributed key escrow model in blockchain with anonymous authentication and committee selection

引用

science China(information sciences) 2023年第3期66卷 289-290页

作者： Axin XIANG Hongfeng GAO Youliang TIAN Liang WAN College of Computer Science and Technology Guizhou University Institute of Cryptography and Data Security Guizhou University Network and Information Management Center Guizhou University State Key Laboratory of Public Big Data Guizhou University Institute of Computer Software and Theory Guizhou University

Dear editor, Key security is of great practical significance and demand to guarantee the security of digital assets in the blockchain system. At present, users prefer to escrow their assets on centralized institutions, but this phenomenon has gone against the unique characteristics of decentralization and anonymity in the blockchain. Among them, the suspense incidents of assets lock or lost are enough to prove that the security of escrowed keys on the exchanges is questionable.

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：