检索结果-内蒙古大学图书馆

Lightweight Task-Oriented Semantic Communication Empowered by Large-Scale AI Models

IEEE Transactions on Vehicular Technology 2025年

作者： Liu, Chuanhong Guo, Caili Yang, Yang Chen, Mingzhe Quek, Tony Q. S. Beijing University of Posts and Telecommunications Beijing Key Laboratory of Network System Architecture and Convergence School of Information and Communication Engineering Beijing100876 China Beijing University of Posts and Telecommunications Beijing Laboratory of Advanced Information Networks School of Information and Communication Engineering Beijing100876 China University of Miami Department of Electrical and Computer Engineering Institute for Data Science and Computing Coral GablesFL United States Singapore University of Technology and Design Dept. of Information Systems Technology and Design 487372 Singapore

Recent studies have focused on leveraging large-scale artificial intelligence (LAI) models to improve semantic representation and compression capabilities. However, the substantial computational demands of LAI models pose significant challenges for real-time communication scenarios. To address this, this paper proposes utilizing knowledge distillation (KD) techniques to extract and condense knowledge from LAI models, effectively reducing model complexity and computation latency. Nevertheless, the inherent complexity of LAI models leads to prolonged inference times during distillation, while their lack of channel awareness compromises the distillation performance. These limitations make standard KD methods unsuitable for task-oriented semantic communication scenarios. To address these issues, we propose a fast distillation method featuring a pre-stored compression mechanism that eliminates the need for repetitive inference, significantly improving efficiency. Furthermore, a channel adaptive module is incorporated to dynamically adjust the transmitted semantic information based on varying channel conditions, enhancing communication reliability and adaptability. In addition, an information bottleneck-based loss function is derived to guide the fast distillation process. Simulation results verify that the proposed scheme outperform baselines in term of task accuracy, model size, computation latency, and training data requirements. © 1967-2012 IEEE.

关键词： NP-hard

来源：评论

学校读者我要写书评

暂无评论

Migrating individuals and probabilistic models on dedas: A comparison on continuous functions

Migrating individuals and probabilistic models on dedas: A c...

引用

10th IASTED International Conference on Artificial Intelligence and Applications, AIA 2010

作者： Muelas, Santiago Mendiburu, Alexander LaTorre, Antonio Peña, José-María Dept. of Computer Systems Architecture and Technology Universidad Politécnica de Madrid Spain Intelligent Systems Group Dept. of Computer Architecture and Technology University of the Basque Country Spain

ISBN: (纸本)9780889868182

One of the most promising areas in which probabilistic graphical models have shown an incipient activity is the field of heuristic optimization and, in particular, in the Estimation of Distribution Algorithms (EDAs). EDAs constitute a well-known family of Evolutionary Computation techniques, similar to Genetic Algorithms. Due to their inherent parallelism, different research lines have been studied trying to improve EDAs from the point of view of execution time and/or accuracy. Among these proposals, we focus on the so-called island-based models. This approach defines several islands (EDA instances) running independently and exchanging information with a given frequency. The information sent by the islands can be a set of individuals or a probabilistic model. This paper presents a comparative study of both information exchanging techniques for a univariate EDA (UMDAg) over a wide set of parameters and problems -the standard benchmark developed for the IEEEWorkshop on Evolutionary Algorithms and other Metaheuristics for Continuous Optimization Problems of the ISDA 2009 Conference. The study concludes that the configurations based on migrating individuals obtain better results.

关键词： Computation theory

来源：评论

学校读者我要写书评

暂无评论

Dynamic assignment with process migration in distributed environments 6th

Dynamic assignment with process migration in distributed env...

引用

6th European Parallel Virtual Machine and Message Passing Interface Users’ Group Meeting, 1999

作者： Czarnul, Pawel Krawczyk, Henryk Dept. of Computer Systems Architecture Technical University of Gdańsk Poland

ISBN: (纸本)3540665498

This paper considers dynamic assignment of application processes to processing nodes. It is assumed that both the application and the system parameters can change dynamically. Monitoring the environment and instrumentation of an application code are used to support the prediction of these parameters. The extension of the PVM environment called DAMPVM1 is proposed which supports both dynamic process allocation and heterogeneous migration. Its suitability for efficient processing of distributed applications is evaluated. © Springer-Verlag Berlin Heidelberg 1999.

关键词： computers

来源：评论

学校读者我要写书评

暂无评论

Using simulation, historical and hybrid estimation systems for enhacing job scheduling on NOWs

Using simulation, historical and hybrid estimation systems f...

引用

2006 IEEE International Conference on Cluster Computing, Cluster 2006

作者： Hanzich, M. Hernández, P. Luque, E. Giné, F. Solsona, F. Lérida, J.L. Dept. Computer Architecture and Operating Systems University Autonoma of Barcelona Spain Dept. Computer Science University of Lleida Spain

ISBN: (纸本)1424403286

The computation capacity of the workstations in an open laboratory is enough to execute not only the local workload but some distributed computation. Unfortunately, the local workload introduces much uncertainty into the predictability of the system, which hinders the applicability of the job scheduling strategies. In this work, we introduce an estimation engine into our job scheduling system, termed CISNE. This prediction capacity allows us guarantee some limits to the turnaround time of parallel jobs. With this aim, three different estimation methods have been proposed and implemented in the CISNE system: a simulation tool, a historical system and an integration of both (hybrid). In this framework, we have compared our proposals to representative estimation methods in the literature. Likewise, we have analyzed these estimation methods in relation to different scheduling policies. These results reveal that the hybrid method achieves the best performance due to the fact that it combines the flexibility of a simulator to represent such a dynamic system as a non-dedicated cluster together with the accuracy given by the historical information. © 2006 IEEE.

关键词： Scheduling

来源：评论

学校读者我要写书评

暂无评论

A CONCEPT OF A FULLY DISTRIBUTED-PROCESSING STRATEGY ORIENTED FOR UNIX LOCAL-AREA NETWORKS

MICROPROCESSING AND MICROPROGRAMMING

引用

MICROPROCESSING AND MICROPROGRAMMING 1994年第10-12期40卷 723-726页

作者： BRUDLO, P KRAWCZYK, H Computer Systems Architecture Dept. Faculty of Electronics Technical University of Gdańsk ul. Narutowicza 11/12 80-952 Gdańsk Poland

A concept of a two levels distributed processing model for local area networks is proposed. The level 1 concerns analysis of current workload of each workstation. Dynamic task distribution in order to achieve load balancing of a whole network is determined on the level 2. To estimate the required parameters two experiments are proposed. They determine relative speeds of workstations and normalized tasks processing time. The proposed approach is the base of the model implementation in a Unix local area network.

关键词： Local area networks

来源：评论

学校读者我要写书评

暂无评论

High performance java sockets for parallel computing on clusters

High performance java sockets for parallel computing on clus...

引用

21st International Parallel and Distributed Processing Symposium, IPDPS 2007

作者： Taboada, Guillermo L. Touriño, Juan Doallo, Ramón Computer Architecture Group Dept. of Electronics and Systems University of A Coruña A Coruña 15071 Spain

ISBN: (纸本)1424409101

The use of Java for parallel programming on clusters relies on the need of efficient communication middleware and high-speed cluster interconnect support. Nevertheless, currently there are no solutions that fully fulfill these issues. In this paper, a Java sockets library has been tailored to increase the efficiency of Java parallel applications on clusters. This library supports high-speed cluster interconnects and its API has been extended to meet the requirements of a high performance Java RMI implementation and Java parallel applications on clusters. Thus, it provides Java with a more efficient communication middleware on clusters. The performance evaluation of this middleware on a Gigabit Ethernet (GbE) and a Scalable Coherent Interface (SCI) cluster has shown experimental evidence of throughput increase. Moreover, qualitative aspects of the solution such as transparency to the user, interoperability with other systems and no need of source code modifications are decisive to boost the performance of existing Java parallel applications and their developments in high performance Java cluster computing. © 2007 IEEE.

关键词： Parallel processing systems

来源：评论

学校读者我要写书评

暂无评论

Applying LSTM for River's Water Level Forecasting as Potential Flood Indication: A Case Study in Greece 15

Applying LSTM for River's Water Level Forecasting as Potenti...

引用

15th International Conference on Information, Intelligence, systems and Applications, IISA 2024

作者： Tziolas, Theodoros Papageorgiou, Elpiniki Papageorgiou, Konstantinos Zardas, Georgios Lekidis, Alexios Panagiotis, Papazoglou University of Thessaly Dept. of Energy Systems Gaiopolis Campus Larisa41500 Greece University of Thessaly Dept. of Electrical and Computer Engineering Volos Greece National and Kapodistrian University of Athens Dept. of Architecture and Computer and Communication Simulation Systems Chalkida Greece

ISBN: (纸本)9798350368833

Floods can have significant life-threatening and economic impacts on human societies. Flood forecasting with data-driven methods is a crucial aspect of disaster risk management, providing valuable insights into potential flood events and allowing for timely mitigation measures. This work illustrates the application of a Long Short-Term Memory (LSTM) neural network for short-term forecasting of a river's water level. The water treatment data used for training the LSTM models were gathered from Sperchios river, located in central Greece. Historical time-series data in a half-hour interval of the river's water level, water velocity, and area precipitation were utilized for the development of the proposed scheme. The developed model is deployed in the backend of a web application for half-hour ahead predictions of the river's water level that are visualized in the frontend. The post-analysis of the deployed forecasting model indicates that it consistently delivers forecasts with minimal error and demonstrates high accuracy in recent flood events. © 2024 IEEE.

关键词： Rivers

来源：评论

学校读者我要写书评

暂无评论

A data parallel formulation of the barnes-hut method for N−body simulations 5th

A data parallel formulation of the barnes-hut method for N−...

引用

5th International Workshop on Applied Parallel Computing, PARA 2000

作者： Amor, M. Argúello, F. López, J. Plata, O. Zapata, E.L. Dept. Electronics and Systems Univ. of La Coruña Spain Dept. Electronics and Computation Univ. of Santiago de Compostela Spain Dept. of Computer Architecture Univ. of Málaga Spain

ISBN: (纸本)354041729X

This paper presents a data-parallel formulation for N−body simulations using the Barnes-Hut method. The tree-structured problem is first linearized by using space-filling curves. This process allows us to use standard data distributions and parallel array operations available in data-parallel languages. A new efficient HPF implementation of the Barnes-Hut method is presented in this paper, characterized by the use of array copy sections to express communications, obtaining efficient point-to-point data transferences. In addition, HPF LOCAL constructs are used to obtain very compact and efficient local (node) codes. © Springer-Verlag Berlin Heidelberg 2001.

关键词： computers

来源：评论

学校读者我要写书评

暂无评论

Photo rendering with swarms: From figurative to abstract pherogenic imaging

Photo rendering with swarms: From figurative to abstract phe...

引用

2013 IEEE Symposium on Computational Intelligence for Creativity and Affective Computing, CICAC 2013 - 2013 IEEE Symposium Series on Computational Intelligence, SSCI 2013

作者： Fernandes, Carlos M. Mora, Antonio Merelo, Juan Julian Cotta, Carlos Rosa, Agostinho C. Dep. of Computer Architecture and Technology University of Granada Spain Systems and Robotics Institute Technical University of Lisbon Portugal Dept. of Computer Languages and Computer Science University of Malaga Spain

ISBN: (纸本)9781467360104

This paper describes the use of ant colony models for photographic rendering, that is, for the depiction or interpretation of photographic images. A description of a previously proposed ant-based edge detection method called pherographia is given in order to contextualize this paper's proposal. Then, the application of the KANTS algorithm to photo-rendering is introduced. KANTS is ant-based clustering algorithm described by a set of equations that model the local behavior of simple units (ants). These units represent data samples. When moving on a heterogeneous 2-dimensional lattice of vectors, global behavior emerges from the local interaction between the ants, which self-organize into clusters. In this paper, a data set consists of the RGB values of a digital (or digitalized) photograph. Two different approaches for the initialization of the environment are tested, each one resulting in radically different output images: when the environment is initialized with the RGB values of the same image that is used to generate the data set, the final result is very similar to the original photo;a random initialization of the environment leads to an abstract interpretation of the photo. The results are contextualized within previous swarm artworks by the authors. © 2013 IEEE.

关键词： Ant colony optimization

来源：评论

学校读者我要写书评

暂无评论

Warp-level divergence in GPUs: Characterization, impact, and mitigation

Warp-level divergence in GPUs: Characterization, impact, and...

引用

20th IEEE International Symposium on High Performance computer architecture, HPCA 2014

作者： Xiang, Ping Yang, Yi Zhou, Huiyang Dept. of Electrical and Computer Engineering North Carolina State University Raleigh NC United States Dept. of Computing Systems Architecture NEC Laboratories America Princeton NJ United States

ISBN: (纸本)9781479930975

High throughput architectures rely on high thread-level parallelism (TLP) to hide execution latencies. In state-of-art graphics processing units (GPUs), threads are organized in a grid of thread blocks (TBs) and each TB contains tens to hundreds of threads. With a TB-level resource management scheme, all the resource required by a TB is allocated/released when it is dispatched to / finished in a streaming multiprocessor (SM). In this paper, we highlight that such TB-level resource management can severely affect the TLP that may be achieved in the hardware. First, different warps in a TB may finish at different times, which we refer to as 'warp-level divergence'. Due to TB-level resource management, the resources allocated to early finished warps are essentially wasted as they need to wait for the longest running warp in the same TB to finish. Second, TB-level management can lead to resource fragmentation. For example, the maximum number of threads to run on an SM in an NVIDIA GTX 480 GPU is 1536. For an application with a TB containing 1024 threads, only 1 TB can run on the SM even though it has sufficient resource for a few hundreds more threads. To overcome these inefficiencies, we propose to allocate and release resources at the warp level. Warps are dispatched to an SM as long as it has sufficient resource for a warp rather than a TB. Furthermore, whenever a warp is completed, its resource is released and can accommodate a new warp. This way, we effectively increase the number of active warps without actually increasing the size of critical resources. We present our lightweight architectural support for our proposed warp-level resource management. The experimental results show that our approach achieves up to 76.0% and an average of 16.0% performance gains and up to 21.7% and an average of 6.7% energy savings at minor hardware overhead. © 2014 IEEE.

关键词： Graphics processing unit

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：