检索结果-内蒙古大学图书馆

IEEE International Conference on Automated Software Engineering (ASE)

作者： Ze Tang Jidong Ge Shangqing Liu Tingwei Zhu Tongtong Xu Liguo Huang Bin Luo National Key Laboratory for Novel Software Technology Nanjing University China Nanyang Technological University Singapore Huawei Software Engineering Application Technology China Department of Computer Science Southern Methodist University USA

Large Language Models (LLMs) have demonstrated remarkable performance in code completion. However, due to the lack of domain-specific knowledge, they may not be optimal in completing code that requires intensive domain knowledge for example completing the library names. Although there are several works that have confirmed the effectiveness of fine-tuning techniques to adapt language models for code completion in specific domains. They are limited by the need for constant fine-tuning of the model when the project is in constant iteration. To address this limitation, in this paper, we propose $k$ NM-LM, a retrieval-augmented language model (R-LM), that integrates domain knowledge into language models without fine-tuning. Different from previous techniques, our approach is able to automatically adapt to different language models and domains. Specifically, it utilizes the in-domain code to build the retrieval-based database decoupled from LM, and then combines it with LM through Bayesian inference to complete the code. The extensive experiments on the completion of intra-project and intra-scenario have confirmed that $k$ NM-LM brings about appreciable enhancements when compared to CodeGPT and UnixCoder. A deep analysis of our tool including the responding speed, storage usage, specific type code completion, and API invocation completion has confirmed that $k$ NM-LM provides satisfactory performance, which renders it highly appropriate for domain adaptive code completion. Furthermore, our approach operates without the requirement for direct access to the language model's parameters. As a result, it can seamlessly integrate with black-box code completion models, making it easy to integrate our approach as a plugin to further enhance the performance of these models.

关键词：

来源：评论

学校读者我要写书评

暂无评论

A Grouping Strategy-Based Progressive Fusion Network for Hyperspectral Image Super-Resolution

A Grouping Strategy-Based Progressive Fusion Network for Hyp...

引用

International Conference on Acoustics, Speech, and Signal Processing (ICASSP)

作者： Guohua Lv Baodong Zhang Yongbiao Gao Guixin Zhao Guotao Wang Juncan Wang Key Laboratory of Computing Power Network and Information Security Ministry of Education Shandong Computer Science Center (National Supercomputer Center in Jinan) Qilu University of Technology (Shandong Academy of Sciences) Jinan China Shandong Provincial Key Laboratory of Computing Power Internet and Service Computing Shandong Fundamental Research Center for Computer Science Jinan China Shandong Key Laboratory of Ubiquitous Intelligent Computing Jinan China Key Laboratory of New Generation Artificial Intelligence Technology and Its Interdisciplinary Application (Southeast University) Ministry of Education China Beihang University Beijing China

ISBN: (数字)9798350368741

ISBN: (纸本)9798350368758

Hyperspectral super-resolution involves combining low-resolution hyperspectral images with high-resolution multispectral images to produce a high-resolution hyperspectral image. Recently, although many methods for hyperspectral image super-resolution have been proposed, they often fail to fully utilize the high similarity among adjacent bands to enhance fusion performance. Therefore, we propose a grouping strategy-based progressive fusion network (GPFNet) for hyperspectral super-resolution. The core of GPFNet is the grouping strategy fusion block (GPF block), in which grouping-based spatial-spectral information fusion and spatial information refinement are performed. We design the spatial-spectral information fusion module (SSIFM) based on grouped convolutions to capture the feature differences from adjacent bands. To refine spatial details, we develop the spatial information enhancement module (SpaEM), which leverages the hierarchical features extracted by the multi-scale feature extraction module (MIEM). Additionally, a progressive fusion strategy, which involves using multiple upsampled hyperspectral images and downsampled multispectral images, further preserves spectral integrity and spatial details. Extensive experiments show that GPFNet outperforms state-of-the-art methods both qualitatively and quantitatively.

关键词： Convolution Superresolution Feature extraction Acoustics Data mining Speech processing Hyperspectral imaging

来源：评论

学校读者我要写书评

暂无评论

MyWAL:performance optimization by removing redundant input/output stack in key-value store

引用

Frontiers of Information technology & Electronic Engineering 2023年第7期24卷 980-993页

作者： Xiao ZHANG Mengyu LI Michael NGULUBE Yonghao CHEN Yiping ZHAO School of Computer Science Northwestern Polytechnical UniversityXi'an 710072China MIIT Key Laboratory of Big Data Storage and Management Northwestern Polytechnical UniversityXi'an 710072China National Engineering Laboratory for Integrated Aero-Space-Ground-Ocean Big Data Application Technology Northwestern Polytechnical UniversityXi'an 710072China School of Software Northwestern Polytechnical UniversityXi'an 710072China

Based on a log-structured merge(LSM)tree,the key-value(KV)storage system can provide high reading performance and optimize random writing *** is widely used in modern data storage systems like e-commerce,online analytics,and real-time *** LSM tree stores new KV data in the memory and flushes to disk in *** prevent data loss in memory if there is an unexpected crash,RocksDB appends updating data in the write-ahead log(WAL)before updating the ***,synchronous WAL significantly reduces writing *** this paper,we present a new WAL mechanism named *** directly manages raw devices(or partitions)instead of saving data on a traditional file *** can avoid useless metadata updating and write data sequentially on *** results show that MyWAL can significantly improve the data writing performance of RocksDB compared to the traditional WAL for small KV data on solid-state disks(SSDs),as much as five to eight times *** non-volatile memory express soild-state drives(NVMe SSDs)and non-volatile memory(NVM),MyWAL can improve data writing performance by 10%–30%.Furthermore,the results of YCSB(Yahoo!Cloud Serving Benchmark)show that the latency decreased by 50%compared with SpanDB.

关键词： key-value(KV)store Log-structured merge(LSM)tree Non-volatile memory(NVM) Non-volatile memory express soild-state drive(NVMe SSD) Write-ahead log(WAL)

来源：评论

学校读者我要写书评

暂无评论

Epidemic Prevention System Based on Voice Recognition Combined with Intelligent Recognition of Mask and Helmet 3

Epidemic Prevention System Based on Voice Recognition Combin...

引用

3rd International Conference on Video, Signal and Image Processing, VSIP 2021

作者： Zhou, Huaming Wang, Aili Li, Meixin Zhao, Yinghui Iwahori, Yuji Heilongjiang Province Key Laboratory Of Laser Spectroscopy Technology And Application Harbin University Of Science And Technology Heilongjiang Harbin China Department Of Computer Science Chubu University Aichi487-8501 Japan

ISBN: (纸本)9781450385886

At present, COVID-19 cross-infection is easy to occur in dense places such as elevators. There are no epidemic prevention measures for construction site elevators on the market, and most of them require manual temperature measurement and reminders to wear masks and helmets to avoid the spread of the epidemic. This paper designs an intelligent epidemic prevention system for the elevator ride process in a modern construction site environment, which can achieve non-contact human temperature measurement, mask and helmet recognition and voice call elevator function. The system uses Arduino UNO as the control core, Kendryte K210 as machine vision processing module, non-contact infrared temperature sensor MLX90614, and voice recognition sensor LD3320. The system has the functions of non-contact temperature detection, mask/helmet recognition(YOLOv3) and voice call elevator. Experimental results showed that the recognition accuracy rate of helmet, mask, voice call elevator is 91.5%, 92.0% and 93.0% respectively. The temperature measurement accuracy rate is 0.2ĝ., which can effectively prevent the spread of the epidemic caused by contact and breathing, and has the advantages of stable, intelligent, and safe work. © 2021 ACM.

关键词： Temperature measurement

来源：评论

学校读者我要写书评

暂无评论

Research on Task Scheduling in Distributed Vulnerability Scanning System 10th

Research on Task Scheduling in Distributed Vulnerability Sca...

引用

10th International Conference on computer Engineering and Networks, CENet 2020

作者： Jiang, Jie Tang, Sixin College of Computer Science and Technology Hengyang Normal University Hengyang421002 China Hunan Provincial Key Laboratory of Intelligent Information Processing and Application Hengyang421002 China

ISBN: (纸本)9789811584619

Network threats caused by system vulnerabilities is increasing gradually, the distributed vulnerability scanning system can scan large-scale and complex networks and report the vulnerability information. Task scheduling is one of the core components in a distributed system. In this paper, we use dynamic optimization algorithms to improve the task scheduling efficiency of distributed vulnerability scanning system, we propose a PSO-based task scheduling scheme and improves the search ability of particles by adjusting algorithm parameters. We compared the time consume when using existing ‘Resource Aware Scheduling algorithm (RASA) with the basic particle swarm optimization (PSO) algorithm and the improved particle swarm optimization (IPSO) algorithm. Our results show that IPSO has better performance than other scheduling methods. © 2021, The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

关键词： Scheduling algorithms

来源：评论

学校读者我要写书评

暂无评论

Multi-Scale Automated Self-Supervised Learning for Recommendation

Multi-Scale Automated Self-Supervised Learning for Recommend...

引用

Intelligent Management and Software Engineering (IMSE), International Conference on

作者： Yuhang Yan Liqin Tian Yujie Gao School of Computer Science Qinghai Normal University Xining China The State Key Laboratory of Tibetan Intelligent Information Processing and Application Xining China School of Computer North China Institute of Science and Technology Yanjiao China

Automated self-supervised learning frameworks have shown significant results in the recommendation domain. However, existing methods such as AutoCF use graph convolutional networks and graph attention mechanisms that are limited to a single expression granularity and cannot fully capture the multi-granularity of user interests and product attributes. To address this problem, we propose MS-AutoCF, a multi-scale automated self-supervised learning recommender system that obtains user and product representations at different granularities through multi-layer graph convolution to match the global and local details of their interest preferences. Meanwhile, we design a query key value-based multi-head graph attention module that can adaptively aggregate and enhance the representations with different granularities. Experimental results show that the MS- AutoCF system achieves significant improvements over existing single-scale methods on several public datasets.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Personalized Learning Path Recommendation Driven by Multi-Modal Knowledge Graphs and Large Models Collaboration

Personalized Learning Path Recommendation Driven by Multi-Mo...

引用

Information technology, Communication Ecosystem and Management (ITCEM), International Conference on

作者： Changhua Zhao Yimei Yang Yujun Yang Bo Tang Chunqiao Mi College of Computer and Artificial Intelligence(Software College) Huaihua University Huaihua China Hunan Provincial Key Laboratory of Ecological Agriculture Intelligent Control Technology Huaihua China Key Laboratory of Wuling-Mountain Health Big Data Intelligent Processing and Application in Hunan Province Universities Huaihua Hunan China

ISBN: (数字)9798331542696

ISBN: (纸本)9798331542702

This study examines ways to create personalized learning pathways that are influenced by large collaborative models and multi-modal knowledge graphs. By integrating multi-modal data to construct a comprehensive knowledge network, this study utilizes the generative and comprehension capabilities of large models to achieve personalized customization of learning resources. The focus of the research is on accurately depicting learner characteristics through multi-modal profiling techniques, ensuring the transparency and credibility of the content generated by large models. The personalized learning route suggestion model proposed in this paper includes two parts: recommendations for learning paths within courses and recommendations for learning paths within learning tasks, aiming to enhance learning efficiency and outcomes through the collaborative effects of multi-modal knowledge graphs and large models. Experimental results indicate that the learning path recommendation model driven by multi-modal knowledge graphs and large collaborative models performs excellently in improving learning outcomes and recommendation performance.

关键词： Knowledge engineering Analytical models Accuracy Biological system modeling Education Ecosystems Collaboration Knowledge graphs Data models Information technology

来源：评论

学校读者我要写书评

暂无评论

Research on Crop Disease and Pest Surveillance based on Big Data and Image Processing Technologies

Research on Crop Disease and Pest Surveillance based on Big ...

引用

Information technology, Communication Ecosystem and Management (ITCEM), International Conference on

作者： Yimei Yang Yujun Yang Huixia Shu Changhua Zhao Shaowei Deng College of Computer and Artificial Intelligence(Software College) Huaihua University Huaihua China Hunan Provincial Key Laboratory of Ecological Agriculture Intelligent Control Technology Huaihua China Key Laboratory of Wuling-Mountain Health Big Data Intelligent Processing and Application in Hunan Province Universities Huaihua Hunan China

ISBN: (数字)9798331542696

ISBN: (纸本)9798331542702

Recently, with the increasing complexity of agricultural systems and climate change, agricultural pest and disease surveillance and early detection have become critical tasks for guaranteeing food stability and promoting sustainable farming growth. Agricultural pest and disease surveillance is a key component of agricultural production, directly affecting crop yield and quality. As big data advances and image processing technologies rapidly, their application in crop pest and disease surveillance has become increasingly widespread. By integrating multi-source data, such as remote sensing data, meteorological data, ecological data, and pest and disease survey data, researchers can build more accurate and efficient pest and disease surveillance models. This paper reviews the research progress of big data and image processing technologies in crop pest and disease surveillance, analyzes the advantages and challenges of current technologies, and proposes future research directions to enhance the automation and intelligence of systems. Additionally, it discusses the integration of deep learning and other techniques to achieve efficient information mining and accurate retrieval of disaster-related data, aiming to provide scientific guidance and technical support for agricultural production.

关键词： Crops Plant diseases Climate change Pest control Agriculture Big Data Surveillance Image processing

来源：评论

学校读者我要写书评

暂无评论

Multi-Domains Personalized Local Differential Privacy Frequency Estimation Mechanism for Utility Optimization

SSRN

引用

SSRN 2024年

作者： Li, Yunfei Fu, Xiaodong Liu, Li Ding, Jiaman Peng, Wei Jia, Lianyin Faculty of Information Engineering and Automation Kunming University of Science and Technology Kunming650500 China Graduate School Yunnan University of Finance and Economics Kunming650221 China Yunnan Provincial Key Laboratory of Computer Technology Application Kunming University of Science and Technology Kunming650500 China

Local Differential Privacy (LDP) has garnered considerable attention in recent years because it does not rely on trusted third parties and has low interactivity and high operational efficiency. However, current LDP frequency estimation mechanisms aggregate data using different privacy budgets within the same data domain, overlooking the aggregation requirements across different data domains. This limits the potential for enhancing the data utility under fixed privacy budgets and meeting user preferences in multiple data domains and privacy budgets. To address this issue, we define a Multi-Domains Personalized Local Differential Privacy (MDPLDP) model that allows users to freely choose data domains and privacy budgets according to their privacy preferences. Furthermore, based on the MDPLDP model, two new frequency estimation mechanisms are proposed: MDPLDP-Generalized Randomized Response and MDPLDP-basic Randomized Aggregatable Privacy-Preserving Ordinal Response. These mechanisms support cross-domains data aggregation and optimize data utility by adjusting the data domains and increasing privacy budgets. Theoretical analysis reveals that these new mechanisms have lower estimation errors than the traditional LDP mechanisms. Experiments on real and synthetic datasets demonstrate that the proposed mechanisms effectively reduce estimation errors and enhance the utility of data-frequency estimation. © 2024, The Authors. All rights reserved.

关键词： Frequency estimation

来源：评论

学校读者我要写书评

暂无评论

Power Analysis Attack on a Lightweight Block Cipher GIFT 9th

Power Analysis Attack on a Lightweight Block Cipher GIFT

引用

9th International Conference on computer Engineering and Networks, CENet2019

作者： Zhang, Jian Li, Lang Li, Qiuping Zhao, Junxia Liang, Xiaoman College of Computer Science and Technology Hengyang Normal University Hengyang421002 China Hunan Provincial Key Laboratory of Intelligent Information Processing and Application Hengyang421002 China

ISBN: (纸本)9789811537523

GIFT is a new lightweight block cipher with smaller area and higher efficiency, which is very suitable for the Internet of Things (IoT) devices with constrained resources. The power analysis attack is an efficient method to extract the key from the cryptographic equipment. However, it is not easy to reveal the key by means of power analysis attack, when the cipher is implemented by hardware. In this article, we present the method of power analysis attack against GIFT. Firstly, we implemented GIFT on FPGA using the SAKURA-G board. Then, we explored the impact of point of interest (POI) on power analysis attack. We proposed the method of power analysis attack against the diffusion layer of GIFT. The experimental results show that the result of power analysis attack is affected by POI, and the key can be recovered when POI is registered. We can reveal the key using correlation power analysis, when targeting the diffusion layer of GIFT. © 2021, Springer Nature Singapore Pte Ltd.

关键词： Side channel attack

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：