检索结果-内蒙古大学图书馆

RE-SEGNN:recurrent semantic evidence-aware graph neural network for temporal knowledge graph forecasting

Science China(Information Sciences) 2025年第2期68卷 103-119页

作者： Wenyu CAI Mengfan LI Xuanhua SHI Yuanxin FAN Quntao ZHU Hai JIN National Engineering Research Center for Big Data Technology and System Services Computing Technology and System LabCluster and Grid Computing LabSchool of Computer Science and TechnologyHuazhong University of Science and Technology

Temporal knowledge graph(TKG) reasoning, has seen widespread use for modeling real-world events, particularly in extrapolation settings. Nevertheless, most previous studies are embedded models, which require both entity and relation embedding to make predictions, ignoring the semantic correlations among different entities and relations within the same timestamp. This can lead to random and nonsensical predictions when unseen entities or relations occur. Furthermore, many existing models exhibit limitations in handling highly correlated historical facts with extensive temporal depth. They often either overlook such facts or overly accentuate the relationships between recurring past occurrences and their current counterparts. Due to the dynamic nature of TKG, effectively capturing the evolving semantics between different timestamps can be *** address these shortcomings, we propose the recurrent semantic evidenceaware graph neural network(RE-SEGNN), a novel graph neural network that can learn the semantics of entities and relations simultaneously. For the former challenge, our model can predict a possible answer to missing quadruples based on semantics when facing unseen entities or relations. For the latter problem, based on an obvious established force, both the recency and frequency of semantic history tend to confer a higher reference value for the current. We use the Hawkes process to compute the semantic trend, which allows the semantics of recent facts to gain more attention than those of distant facts. Experimental results show that RE-SEGNN outperforms all SOTA models in entity prediction on 6 widely used datasets, and 5 datasets in relation prediction. Furthermore, the case study shows how our model can deal with unseen entities and relations.

关键词： knowledge graph reasoning temporal knowledge graph Hawkes process semantic evidence

来源：评论

学校读者我要写书评

暂无评论

Precise control of page cache for containers

引用

Frontiers of Computer Science 2024年第2期18卷 17-30页

作者： Kun WANG Song WU Shengbang LI Zhuo HUANG Hao FAN Chen YU Hai JIN National Engineering Research Center for Big Data Technology and System Services Computing Technology and System LabCluster and Grid Computing Lab School of Computer Science and TechnologyHuazhong University of Science and TechnologyWuhan 430074China

Container-based virtualization isbecoming increasingly popular in cloud computing due to its efficiency and *** isolation is a fundamental property of *** works have indicated weak resource isolation could cause significant performance degradation for containerized applications and enhanced resource ***,current studies have almost not discussed the isolation problems of page cache which is a key resource for *** leverage memory cgroup to control page cache ***,existing policy introduces two major problems in a container-based ***,containers can utilize more memory than limited by their cgroup,effectively breaking memory ***,the Os kernel has to evict page cache to make space for newly-arrived memory requests,slowing down containerized *** paper performs an empirical study of these problems and demonstrates the performance impacts on containerized *** we propose pCache(precise control of page cache)to address the problems by dividing page cache into private and shared and controlling both kinds of page cache separately and *** do so,pCache leverages two new technologies:fair account(f-account)and evict on demand(EoD).F-account splits the shared page cache charging based on per-container share to prevent containers from using memory for free,enhancing memory *** EoD reduces unnecessary page cache evictions to avoid the performance *** evaluation results demonstrate that our system can effectively enhance memory isolation for containers and achieve substantial performance improvement over the original page cache management policy.

关键词： page cache memory cgroup container isolation cloud computing

来源：评论

学校读者我要写书评

暂无评论

Answering reachability queries with ordered label constraints over labeled graphs

引用

Frontiers of Computer Science 2024年第1期18卷 105-117页

作者： Daoliang HE Pingpeng YUAN Hai JIN National Engineering Research Center for Big Data Technology and System Services Computing Technology and System LabCluster and Grid Computing LabSchool of Computer Science&TechnologyHuazhong University of Science and TechnologyWuhan 430074China

Reachability query plays a vital role in many graph analysis *** researches proposed many methods to efficiently answer reachability queries between vertex *** many real graphs are labeled graph,it highly demands Label-Constrained Reachability(LCR)query inwhich constraint includes a set of labels besides vertex *** researches proposed several methods for answering some LCR queries which require appearance of some labels specified in constraints in the *** that constraint may be a label set,query constraint may be ordered labels,namely OLCR(Ordered-Label-Constrained Reachability)queries which retrieve paths matching a sequence of ***,no solutions are available for ***,we propose DHL,a novel bloom filter based indexing technique for answering OLCR *** can be used to check reachability between vertex *** the answers are not no,then constrained DFS is ***,we employ DHL followed by performing constrained DFS to answer OLCR *** show that DHL has a bounded false positive rate,and it's powerful in saving indexing time and *** experiments on 10 real-life graphs and 12 synthetic graphs demonstrate that DHL achieves about 4.8-22.5 times smaller index space and 4.6-114 times less index construction time than two state-of-art techniques for LCR queries,while achieving comparable query response *** results also show that our algorithm can answer OLCR queries effectively.

关键词： graph ordered-label-constrained reachability partial index bloom filter query processing

来源：评论

学校读者我要写书评

暂无评论

Instance segmentation of high-resolution remote sensing images based on Bayesian classifier and discrete cosine transform 24

Instance segmentation of high-resolution remote sensing imag...

引用

4th International Conference on Artificial Intelligence, Automation and High Performance computing, AIAHPC 2024

作者： He, Jiangtao Wang, Licai Yu, Jintao You, Peihai Huang, Yixuan Big data research and development center North China Institute of Computing Technology China

ISBN: (纸本)9798400710049

The popularization and widespread application of high-resolution remote sensing imagery have made Instance Segmentation a hot research area. The difficulty of improving the instance segmentation accuracy of high-precision fine-grained remote sensing images has always been a research challenge. This paper proposes a new model, Oriented Bayes DCT, which effectively improves the accuracy of instance segmentation. The model utilizes a region proposal network to encode features, generate proposals, align the candidate boxes, and then uses Bayesian methods to predict class and box offsets. Compared to traditional methods of generating candidate boxes like RPN or Oriented RPN, the candidate boxes obtained using the Naive Bayes method are more accurate. Additionally, the model employs a mask prediction branch to predict segmentation. Through Discrete Cosine Transform (DCT), high-resolution binary grid masks are encoded into compact vectors, achieving precise instance segmentation of fine-grained remote sensing images. Compared to predicting the mask part using the Mask Head method, using the discrete cosine transform for encoding and decoding the ground truth masks results in lower error rates. Without tricks, Oriented Bayes DCT with ResNet50 achieves state-of-the-art segmentation accuracy on a commonly-used datasets iSAID for instance segmentation (37.7% mAP) and ResNet101 (38.5% mAP). © 2024 Copyright held by the owner/author(s).

关键词： Image segmentation

来源：评论

学校读者我要写书评

暂无评论

UCat: heterogeneous memory management for unikernels

引用

Frontiers of Computer Science 2023年第1期17卷 51-61页

作者： Chong TIAN Haikun LIU Xiaofei LIAO Hai JIN National Engineering Research Center for Big Data Technology and System Services Computing Technology and System Lab/Cluster and Grid Computing LabSchool of Computing Science and TechnologyHuazhong University of Science and TechnologyWuhan 430074China

Unikernels provide an efficient and lightweight way to deploy cloud computing services in application-specialized and single-address-space virtual machines (VMs). They can efficiently deploy hundreds of unikernel-based VMs in a single physical server. In such a cloud computing platform, main memory is the primary bottleneck resource for high-density application deployment. Recently, non-volatile memory (NVM) technologies has become increasingly popular in cloud data centers because they can offer extremely large memory capacity at a low expense. However, there still remain many challenges to utilize NVMs for unikernel-based VMs, such as the difficulty of heterogeneous memory allocation and high performance overhead of address *** this paper, we present UCat, a heterogeneous memory management mechanism that support multi-grained memory allocation for unikernels. We propose front-end/back-end cooperative address space mapping to expose the host memory heterogeneity to unikernels. UCat exploits large pages to reduce the cost of two-layer address translation in virtualization environments, and leverages slab allocation to reduce memory waste due to internal memory fragmentation. We implement UCat based on a popular unikernel--OSv and conduct extensive experiments to evaluate its efficiency. Experimental results show that UCat can reduce the memory consumption of unikernels by 50% and TLB miss rate by 41%, and improve the throughput of real-world benchmarks such as memslap and YCSB by up to 18.5% and 14.8%, respectively.

关键词： unikernel virtualization non-volatile memory heterogeneous memory large page slab allocation

来源：评论

学校读者我要写书评

暂无评论

research on large model table question answering method based on thought of chain and vector matching filtering 3

Research on large model table question answering method base...

引用

3rd International Conference on Electronic Information Engineering and data Processing, EIEDP 2024

作者： Tang, Haichao Wang, Licai Hu, Xun Luo, Qibin Big data research and development center North China Institute of Computing Technology Beijing China

ISBN: (纸本)9781510680531

Implementing table-based question answering can fully leverage the rich information in tabular data, and the development of large language models has provided strong natural language processing and understanding capabilities. Current research on table-based question answering with large language models mainly focuses on how to improve text-to-SQL abilities and the development of table pre-training models, lacking in studies on feasible technical approaches for applying general large language models to engineering applications involving large-scale database table question answering. In response, combining techniques such as prompt learning and chain of thought,this paper constructs a database question answering chain of thought based on large language model, proposes a table filtering scheme based on vectorized matching of table structural information, specifically addressing poor table filtering effects, and constructs a database optimization chain of thought in the context of de-identification scenarios. Experimental verification shows that after adopting the chain of thought construction and vector matching filtering scheme, the database question answering capability of the large model has been effectively improved. © 2024 SPIE.

关键词： Information filtering

来源：评论

学校读者我要写书评

暂无评论

Legal Event Detection Model Based on ERNIE 3.0-CRF and Ensemble Learning 5

Legal Event Detection Model Based on ERNIE 3.0-CRF and Ensem...

引用

5th International Seminar on Artificial Intelligence, Networking and Information Technology, AINIT 2024

作者： Wang, Siyu Wang, Licai Zhang, Ke North China Institute of Computing Technology Big Data Research and Development Center Beijing China

ISBN: (纸本)9798350385557

Recognizing events is a prerequisite for making judgments, so discovering events in legal documents is important for the analysis of legal case. Legal Event Detection aims to extract legal event trigger words from legal documents and categorize events and is the basis of many judicial AI applications. Oriented to the needs of understanding and analyzing legal events in the judicial trial process, and in response to the low accuracy of event detection caused by the many types of events and little sample data in the legal domain, this paper constructs a pre-training model oriented to event detection in the legal domain, which makes full use of ERNIE 3.0's powerful semantic representation learning capability and CRF's effective modeling capability for sequence annotation tasks. In addition, we further enhance the effectiveness of the model through data augmentation and ensemble learning. The experimental results show that the method works better compared to the baseline model and improves the prediction accuracy of legal event detection. © 2024 IEEE.

关键词： Learning systems

来源：评论

学校读者我要写书评

暂无评论

ARCHER:a ReRAM-based accelerator for compressed recommendation systems

引用

Frontiers of Computer Science 2024年第5期18卷 147-160页

作者： Xinyang SHEN Xiaofei LIAO Long ZHENG Yu HUANG Dan CHEN Hai JIN National Engineering Research Center for Big Data Technology and System Services Computing Technology and System LabClusters and Grid Computing LabSchool of Computer Science and TechnologyHuazhong University of Science and TechnologyWuhan 430074China

Modern recommendation systems are widely used in modern data *** random and sparse embedding lookup operations are the main performance bottleneck for processing recommendation systems on traditional platforms as they induce abundant data movements between computing units and ***-based processing-in-memory(PIM)can resolve this problem by processing embedding vectors where they are ***,the embedding table can easily exceed the capacity limit of a monolithic ReRAM-based PIM chip,which induces off-chip accesses that may offset the PIM ***,we deploy the decomposed model on-chip and leverage the high computing efficiency of ReRAM to compensate for the decompression performance *** this paper,we propose ARCHER,a ReRAM-based PIM architecture that implements fully yon-chip recommendations under resource ***,we make a full analysis of the computation pattern and access pattern on the decomposed *** on the computation pattern,we unify the operations of each layer of the decomposed model in multiply-and-accumulate *** on the access observation,we propose a hierarchical mapping schema and a specialized hardware design to maximize resource *** the unified computation and mapping strategy,we can coordinatethe inter-processing elements *** evaluation shows that ARCHER outperforms the state-of-the-art GPU-based DLRM system,the state-of-the-art near-memory processing recommendation system RecNMP,and the ReRAM-based recommendation accelerator REREC by 15.79×,2.21×,and 1.21× in terms of performance and 56.06×,6.45×,and 1.71× in terms of energy savings,respectively.

关键词： recommendation system ReRAM processing-in-memory embedding layer

来源：评论

学校读者我要写书评

暂无评论

RIF-Proto: A Few-Shot Relation Extraction for Prototype Networks Based on Relational Information Features 5

RIF-Proto: A Few-Shot Relation Extraction for Prototype Netw...

引用

5th International Seminar on Artificial Intelligence, Networking and Information Technology, AINIT 2024

作者： Zhang, Shuai Wang, Licai Huang, Yangchen North China Institute of Computing Technology Big Data Research and Development Center Beijing China

ISBN: (纸本)9798350385557

Relation extraction stands as a core task within the realm of information extraction, focusing on identifying the target relations between entities in text. While traditional relation extraction has been extensively researched, certain relations lack sufficient labeled data, rendering traditional models ineffective in practical research. Hence, the research focus in this field has shifted towards few-shot learning methods that leverage limited labeled data for training. Significant progress has been made in few-shot relation extraction, with prototype networks garnering attention from scholars for their outstanding performance in few-shot tasks. Prototype networks utilize the clustering idea to achieve relationship classification simply and efficiently, and their training process is mainly divided into three parts: generation of relationship prototypes, relationship classification, and loss calculation. The majority of current research is centered around leveraging external information to enhance the generation of prototypes and loss functions, but due to the difficulty of obtaining external data and the high complexity of data processing, it greatly increases the time cost and the difficulty of implementation. Given the aforementioned challenges, this paper introduces a method for prototype refinement leveraging relationship similarity among existing data. This approach aims to effectively address the issue of significant deviations in relationship prototypes. Initially, extract relationship vectors from the instances in both the support set and query set, followed by computing the relationship similarity between samples based on these vectors, and finally, this paper constructs a prototype correction module, which utilizes the relationship similarity to assign weights to the samples in the support set, to achieve the purpose of prototype correction. In the experiments against the standard dataset FewRel1.0., we observe that the method substantially improves the ex

关键词： Query processing

来源：评论

学校读者我要写书评

暂无评论

Large Language Model Based on Full-Text Retrieval for Temporal Knowledge Q&A Approach 5

Large Language Model Based on Full-Text Retrieval for Tempor...

引用

5th International Seminar on Artificial Intelligence, Networking and Information Technology, AINIT 2024

作者： Li, Zhidong Wang, Licai Luo, Qibin Qiao, Silong North China Institute of Computing Technology Big Data Research and Development Center Beijing China

ISBN: (纸本)9798350385557

Knowledge Q&A is one of the hot research topics in the field of natural language processing, and temporal knowledge Q&A is a difficult area of Q&A reasoning because it also needs to consider the temporal relationship of knowledge. Today's research usually focuses on the word vector similarity between knowledge and questions as an important basis for answering, while ignoring the sentence granularity semantic information embedded in the knowledge. In this paper, we propose a method of temporal knowledge Q&A for large language models based on full-text retrieval, firstly, the sentence granularity knowledge recall is performed by Elasticsearch so that large language models can learn the knowledge that is highly relevant to the problem, and then verify the temporal knowledge Q&A ability of large language models under Zero-shot, One-shot and Few-shot. The experiments were conducted on the ICEWS05-15 dataset, and the accuracy of answers was significantly improved, demonstrating the effectiveness of the temporal knowledge Q&A method for large language models based on Elasticsearch. © 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：