检索结果-内蒙古大学图书馆

How far are we to GPT-4V?Closing the gap to commercial multimodal models with open-source suites

science China(Information sciences) 2024年第12期67卷 5-22页

作者： Zhe CHEN Weiyun WANG Hao TIAN Shenglong YE Zhangwei GAO Erfei CUI Wenwen TONG Kongzhi HU Jiapeng LUO Zheng MA Ji MA Jiaqi WANG Xiaoyi DONG Hang YAN Hewei GUO Conghui HE Botian SHI Zhenjiang JIN Chao XU Bin WANG Xingjian WEI Wei LI Wenjian ZHANG Bo ZHANG Pinlong CAI Licheng WEN Xiangchao YAN Min DOU Lewei LU Xizhou ZHU Tong LU Dahua LIN Yu QIAO Jifeng DAI Wenhai WANG State Key Laboratory for Novel Software Technology Nanjing University Shanghai AI Laboratory School of Computer Science Fudan University SenseTime Research Department of Information Engineering The Chinese University of Hong Kong Department of Electronic Engineering Tsinghua University

In this paper, we introduce InternVL 1.5, an open-source multimodal large language model(MLLM) to bridge the capability gap between open-source and proprietary commercial models in multimodal understanding. We introduce three simple improvements.(1) Strong vision encoder: we explored a continuous learning strategy for the large-scale vision foundation model — InternViT-6B, boosting its visual understanding capabilities, and making it can be transferred and reused in different LLMs.(2) Dynamic high-resolution: we divide images into tiles ranging from 1 to 40 of 448×448 pixels according to the aspect ratio and resolution of the input images, which supports up to 4K resolution input.(3) High-quality bilingual dataset: we carefully collected a high-quality bilingual dataset that covers common scenes, document images,and annotated them with English and Chinese question-answer pairs, significantly enhancing performance in optical character recognition(OCR) and Chinese-related tasks. We evaluate InternVL 1.5 through a series of benchmarks and comparative studies. Compared to both open-source and proprietary commercial models, InternVL 1.5 shows competitive performance, achieving state-of-the-art results in 8 of 18 multimodal benchmarks. Code and models are available at https://***/OpenGVLab/InternVL.

关键词： multimodal model open-source vision encoder dynamic resolution bilingual dataset

来源：评论

学校读者我要写书评

暂无评论

Fake News Encoder Classifier (FNEC) for Online Published News Related to COVID-19 Vaccines

引用

Intelligent Automation & Soft Computing 2023年第7期37卷 73-90页

作者： Asma Qaiser Saman Hina Abdul Karim Kazi Saad Ahmed Raheela Asif Department of Computer Science and Information Technology NED University of Engineering and TechnologyKarachi75270Pakistan Department of Computer Science Iqra UniversityKarachi76400Pakistan Department of Software Engineering NED University of Engineering and TechnologyKarachi75270Pakistan

In the past few years,social media and online news platforms have played an essential role in distributing news content *** of the authenticity of news has become a major *** the COVID-19 outbreak,misinformation and fake news were major sources of confusion and insecurity among the general *** the first quarter of the year 2020,around 800 people died due to fake news relevant to *** major goal of this research was to discover the best learning model for achieving high accuracy and performance.A novel case study of the Fake News Classification using ELECTRA model,which achieved 85.11%accuracy score,is thus reported in this *** addition to that,a new novel dataset called COVAX-Reality containing COVID-19 vaccine-related news has been *** the COVAX-Reality dataset,the performance of FNEC is compared to several traditional learning models i.e.,Support Vector Machine(SVM),Naive Bayes(NB),Passive Aggressive Classifier(PAC),Long Short-Term Memory(LSTM),Bi-directional LSTM(Bi-LSTM)and Bi-directional Encoder Representations from Transformers(BERT).For the evaluation of FNEC,standard metrics(Precision,Recall,Accuracy,and F1-Score)were utilized.

关键词： Deep learning fake news detection machine learning transformer model classification

来源：评论

学校读者我要写书评

暂无评论

Characterization of Memory Access in Deep Learning and Its Implications in Memory Management

引用

computers, Materials & Continua 2023年第7期76卷 607-629页

作者： Jeongha Lee Hyokyung Bahn Institute of Artificial Intelligence and Software Ewha UniversitySeoul03760Korea Department of Computer Science and Engineering Ewha UniversitySeoul03760Korea

Due to the recent trend of software intelligence in the Fourth Industrial Revolution,deep learning has become a mainstream workload for modern computer *** the data size of deep learning increasingly grows,managing the limited memory capacity efficiently for deep learning workloads becomes *** this paper,we analyze memory accesses in deep learning workloads and find out some unique characteristics differentiated from traditional ***,when comparing instruction and data accesses,data access accounts for 96%–99%of total memory accesses in deep learning workloads,which is quite different from traditional ***,when comparing read and write accesses,write access dominates,accounting for 64%–80%of total memory ***,although write access makes up the majority of memory accesses,it shows a low access bias of 0.3 in the Zipf ***,in predicting re-access,recency is important in read access,but frequency provides more accurate information in write *** on these observations,we introduce a Non-Volatile Random Access Memory(NVRAM)-accelerated memory architecture for deep learning workloads,and present a new memory management policy for this *** considering the memory access characteristics of deep learning workloads,the proposed policy improves memory performance by 64.3%on average compared to the CLOCK policy.

关键词： Memory access deep learning machine learning memory access memory management CLOCK

来源：评论

学校读者我要写书评

暂无评论

Accelerating Distributed Urban Traffic Simulation via Enhanced Stale Synchronous Parallelism 11

Accelerating Distributed Urban Traffic Simulation via Enhanc...

引用

11th IEEE International Conference on Behavioural and Social Computing, BESC 2024

作者： Zhu, Haojia Ma, Ran Jin, Jiahui Lu, Hongru School of Computer Science and Engineering Southeast University China School of Software Engineering Southeast University China School of Computer Science Nanjing Audit University Nanjing China

ISBN: (纸本)9798331531904

Modeling urban mobility behaviours with micro-scopic traffic flow simulation is now crucial for studying intel-ligent urban decision-making algorithms, such as traffic light control and road congestion charging. However, in urban-scale traffic environments, simulation is computationally intensive. Most existing studies use distributed traffic simulations based on the Bulk Synchronous Parallel (BSP) model, leading to significant time and data transfer costs due to the need for coordination and data synchronization. To address these issues, we propose an enhanced Stale Synchronous Parallel (ESSP) model. Our model reduces the waiting time of the simulation compute nodes, and we design it to control and measure errors during simulation. We also develop a distributed traffic simulator to validate our model, and the results show a performance improvement of 30% to 50% compared to conventional synchronization methods. © 2024 IEEE.

关键词： Traffic congestion

来源：评论

学校读者我要写书评

暂无评论

Efficient Federated Learning Using Dynamic Update and Adaptive Pruning with Momentum on Shared Server Data

引用

ACM Transactions on Intelligent Systems and Technology 2024年第6期15卷 1-28页

作者： Liu, Ji Jia, Juncheng Zhang, Hong Yun, Yuhui Wang, Leye Zhou, Yang Dai, Huaiyu Dou, Dejing Hithink RoyalFlush Information Network Co. Ltd. Hangzhou China School of Computer Science and Technology Soochow University Suzhou China Collaborative Innovation Center of Novel Software Technology and Industrialization Nanjing China Baidu Research Beijing China Key Lab of High Confidence Software Technologies Ministry of Education and Software Institute Peking University Beijing China Department of Computer Science and Software Engineering Auburn University AuburnAL United States Department of Electrical and Computer Engineering North Carolina State University RaleighNC United States Bedi Cloud and School of Computer Science Fudan University Beijing China

Despite achieving remarkable performance, Federated Learning (FL) encounters two important problems, i.e., low training efficiency and limited computational resources. In this article, we propose a new FL framework, i.e., FedDUMAP, with three original contributions, to leverage the shared insensitive data on the server in addition to the distributed data in edge devices so as to efficiently train a global model. First, we propose a simple dynamic server update algorithm, which takes advantage of the shared insensitive data on the server while dynamically adjusting the update steps on the server in order to speed up the convergence and improve the accuracy. Second, we propose an adaptive optimization method with the dynamic server update algorithm to exploit the global momentum on the server and each local device for superior accuracy. Third, we develop a layer-adaptive model pruning method to carry out specific pruning operations, which is adapted to the diverse features of each layer so as to attain an excellent tradeoff between effectiveness and efficiency. Our proposed FL model, FedDUMAP, combines the three original techniques and has a significantly better performance compared with baseline approaches in terms of efficiency (up to 16.9 times faster), accuracy (up to 20.4% higher), and computational cost (up to 62.6% smaller). © 2024 Copyright held by the owner/author(s). Publication rights licensed to ACM.

关键词： Federated learning

来源：评论

学校读者我要写书评

暂无评论

Suitable and Style-Consistent Multi-Texture Recommendation for Cartoon Illustrations

引用

ACM Transactions on Multimedia Computing, Communications and Applications 2024年第7期20卷 1-26页

作者： Wu, Huisi Wang, Zhaoze Li, Yifan Liu, Xueting Lee, Tong-Yee College of Computer Science and Software Engineering Shenzhen University No. 3688 Nanhai Road Nanshan District Guangdong Shenzhen518060 China Dept. of Computer Science and Information Engineering National Cheng-Kung University No. 1 University Road Tainan70101 Taiwan

Texture plays an important role in cartoon illustrations to display object materials and enrich visual experiences. Unfortunately, manually designing and drawing an appropriate texture is not easy even for proficient artists, let alone novice or amateur people. While there exist tons of textures on the Internet, it is not easy to pick an appropriate one using traditional text-based search engines. Although several texture pickers have been proposed, they still require the users to browse the textures by themselves, which is still labor-intensive and time-consuming. In this article, an automatic texture recommendation system is proposed for recommending multiple textures to replace a set of user-specified regions in a cartoon illustration with visually pleasant look. Two measurements, the suitability measurement and the style-consistency measurement, are proposed to make sure that the recommended textures are suitable for cartoon illustration and at the same time mutually consistent in style. The suitability is measured based on the synthesizability, cartoonity, and region fitness of textures. The style-consistency is predicted using a learning-based solution since it is subjective to judge whether two textures are consistent in style. An optimization problem is formulated and solved via the genetic algorithm. Our method is validated on various cartoon illustrations, and convincing results are obtained. © 2024 Copyright held by the owner/author(s). Publication rights licensed to ACM.

关键词： Textures

来源：评论

学校读者我要写书评

暂无评论

Zero Trust Networks: Evolution and Application from Concept to Practice

引用

computers, Materials & Continua 2025年第2期82卷 1593-1613页

作者： Yongjun Ren Zhiming Wang Pradip Kumar Sharma Fayez Alqahtani Amr Tolba Jin Wang School of Computer Science School of Cyber Science and EngineeringEngineering Research Center of Digital ForensicsMinistry of EducationNanjing University of Information Science and TechnologyNanjing210044China Department of Computing Science University of AberdeenAberdeenAB243FXUK Software Engineering Department College of Computer and Information SciencesKing Saud UniversityRiyadh12372Saudi Arabia Computer Science Department Community CollegeKing Saud UniversityRiyadh11437Saudi Arabia Sanya Institute of Hunan University of Science and Technology Sanya572024China

In the context of an increasingly severe cybersecurity landscape and the growing complexity of offensive and defen-sive techniques,Zero Trust Networks(ZTN)have emerged as a widely recognized *** Trust not only addresses the shortcomings of traditional perimeter security models but also consistently follows the fundamental principle of“never trust,always verify.”Initially proposed by John Cortez in 2010 and subsequently promoted by Google,the Zero Trust model has become a key approach to addressing the ever-growing security threats in complex network *** paper systematically compares the current mainstream cybersecurity models,thoroughly explores the advantages and limitations of the Zero Trust model,and provides an in-depth review of its components and key ***,it analyzes the latest research achievements in the application of Zero Trust technology across various fields,including network security,6G networks,the Internet of Things(IoT),and cloud computing,in the context of specific use *** paper also discusses the innovative contributions of the Zero Trust model in these fields,the challenges it faces,and proposes corresponding solutions and future research directions.

关键词： Zero trust cybersecurity software-defined perimeter micro-segmentation internet of things

来源：评论

学校读者我要写书评

暂无评论

Cardinality estimation for property graph queries with gated learning approach on the graph database

引用

Multimedia Tools and Applications 2025年第11期84卷 9159-9183页

作者： He, Zhenzhen Yu, Jiong Du, Xusheng Guo, Binglei Li, Ziyang Li, Zhe School of Information Science and Engineering Xinjiang University Urumqi China School of Computer Engineering Hubei University of Arts and Science Xiangyang China School of Software Xinjiang University Urumqi China Department of Electronic and Information Engineering The Hong Kong Polytechnic University Hong Kong

With the increasing complexity of graph query processing tasks, it is difficult for users to obtain the accurate cardinality before or during the execution of query tasks. Accurate estimate query cardinality is crucial for property graph data model, which usually involves entities, multiple joins, and various types of properties (key-value pairs). While learning-based approaches have been used in query optimization, estimating the cardinality of property graph queries is still particularly challenging. In this paper, we first formally represent each property graph query and then divide the estimation process into three phases: query execution, training data generation, and model training. Secondly, we construct a data pool to deal the updating query workloads in the training data generation phase. Thirdly, we utilize the gate mechanism to develop a cardinality estimation framework based on deep learning neural networks for property graph queries in the Neo4j database, and we propose a hybrid loss function to optimize the training process. Finally, we adopt a method of pre-aggregating the underlying data to speed up query execution in the first phase. The result of the experiment on two real-world datasets shows that our optimized methods can effectively improve the estimation accuracy of the deep learning model, and our query estimator outperforms other deep learning models compared in terms of estimation accuracy. © The Author(s), under exclusive licence to Springer science+Business Media, LLC, part of Springer Nature 2024.

关键词： Query processing

来源：评论

学校读者我要写书评

暂无评论

Weakly Supervised Intracranial Hemorrhage Segmentation with YOLO and an Uncertainty Rectified Segment Anything Model 4th

Weakly Supervised Intracranial Hemorrhage Segmentation with...

引用

4th International Workshop on Imaging and Treatment Challenges, SWITCH 2024, and 6th International Challenge on Ischemic Stroke Lesion Segmentation Challenge, ISLES 2024, Held in Conjunction with Medical Image Computing and computer Assisted Intervention, MICCAI 2024

作者： Spiegler, Pascal Rasoulian, Amirhossein Xiao, Yiming Department of Computer Science and Software Engineering Concordia University Montreal Canada NeuroRx Research Montreal Canada

ISBN: (纸本)9783031811005

Intracranial hemorrhage (ICH) is a life-threatening condition that requires rapid and accurate diagnosis to improve treatment outcomes and patient survival rates. Recent advancements in supervised deep learning have greatly improved the analysis of medical images, but often rely on extensive datasets with high-quality annotations, which are costly, time-consuming, and require medical expertise to prepare. To mitigate the need for large amounts of expert-prepared segmentation data, we have developed a novel weakly supervised ICH segmentation method that utilizes the YOLO object detection model and an uncertainty-rectified Segment Anything Model (SAM). In addition, we have proposed a novel point prompt generator for this model to further improve segmentation results with YOLO-predicted bounding box prompts. Our approach achieved a high accuracy of 0.933 and an AUC of 0.796 in ICH detection, along with a mean Dice score of 0.629 for ICH segmentation, outperforming existing weakly supervised and popular supervised (UNet and Swin-UNETR) approaches. Overall, the proposed method provides a robust and accurate alternative to the more commonly used supervised techniques for ICH quantification without requiring refined segmentation ground truths during model training. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

关键词： Image segmentation

来源：评论

学校读者我要写书评

暂无评论

MCM-Llama: A Fine-Tuned Large Language Model for Real-Time Threat Detection through Security Event Correlation 4

MCM-Llama: A Fine-Tuned Large Language Model for Real-Time T...

引用

4th IEEE International Conference on Electrical, computer, and Energy Technologies, ICECET 2024

作者： Diakhame, Mouhamadou Lamine Diallo, Cherif Mejri, Mohamed Gaston Berger University Department of Computer Science Saint Louis Senegal Laval University Department of Computer Science & Software Engineering Quebec Canada

ISBN: (纸本)9798350395914

The correlation of security alerts is crucial for effective threat detection and mitigation in modern cybersecurity landscapes. With the exponential growth in the volume of generated alerts, there arises an imperative to continually develop new, efficient methods for alert correlation. The use of Large Language Models (LLMs) presents new avenues for enhancing real-time threat detection in the field of cybersecurity. In this paper, we introduce MCM-Llama, a fine-tuned language model aimed at improving security event correlation for real-time threat detection. Building upon our prior research, we augment the current framework by integrating MCM-Llama and introducing a novel architecture to practically implement our proposition. In doing so, we replace the correlation module previously based on Named Entity Recognition (NER) and semantic similarity with a fine-tuned Large Language Model (LLM). This transition allows us to capitalize on the advanced capabilities of LLM in comprehending and correlating security events dynamically. Through this enhancement, we aim to significantly improve the accuracy and efficiency of our threat detection system while leveraging the full potential of LLM technology in cybersecurity. Our experimental results validate the efficacy of the proposed approach by significantly improving the correlation rate, thereby reinforcing the pivotal role of Large Language Models (LLMs) in advancing security event correlation and threat detection methodologies. © 2024 IEEE.

关键词： Semantics

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：