检索结果-内蒙古大学图书馆

ACM/IEEE 51st Annual international symposium on computer Architecture (ISCA)

作者： Shivdikar, Kaustubh Agostini, Nicolas Bohm Jayaweera, Malith Jonatan, Gilbert Abellan, Jose L. Joshi, Ajay Kim, John Kaeli, David Northeastern Univ Boston MA 02115 USA Korea Adv Inst Sci & Technol Daejeon South Korea Univ Murcia Murcia Spain Boston Univ Boston MA USA

ISBN: (纸本)9798350326598;9798350326581

Graph Neural networks (GNNs) are emerging as a formidable tool for processing non-euclidean data across various domains, ranging from social network analysis to bioinformatics. Despite their effectiveness, their adoption has not been pervasive because of scalability challenges associated with large-scale graph datasets, particularly when leveraging message passing. They exhibit irregular sparsity patterns, resulting in unbalanced compute resource utilization. Prior accelerators investigating Gustavson's technique adopted look-ahead buffers for prefetching data, aiming to prevent compute stalls. However, these solutions lead to inefficient use of the on-chip memory, leading to redundant data residing in cache. To tackle these challenges, we introduce NeuraChip, a novelGNN spatial accelerator based on Gustavson's algorithm. NeuraChip decouples the multiplication and addition computations in sparse matrix multiplication. This separation allows for independent exploitation of their unique data dependencies, facilitating efficient resource allocation. We introduce a rolling eviction strategy to mitigate data idling in on-chip memory as well as address the prevalent issue of memory bloat in sparse graph computations. Furthermore, the compute resource load balancing is achieved through a dynamic reseeding hash-based mapping, ensuring uniform utilization of computing resources agnostic of sparsity patterns. Finally, we present NeuraSim, an open-source, cycle-accurate, multi-threaded, modular simulator for comprehensive performance analysis. Overall, NeuraChip presents a significant improvement, yielding an average speedup of 22.1x over Intel's MKL, 17.1x over NVIDIA's cuSPARSE, 16.7x over AMD's hipSPARSE, and 1.5x over prior state-of-the-art SpGEMM accelerator and 1.3x over GNN accelerator. The source code for our open-sourced simulator and performance visualizer is publicly accessible on GitHub1.

关键词： Decoupled Computations Graph Neural networks (GNN) Hardware-software co-design On-chip Memory Sparse Matrix Multiplication (SpGEMM) Spatial Accelerators

来源：评论

学校读者我要写书评

暂无评论

Optimal Sizing and Allocation of distributed Energy Resources using Genetic Algorithms: Case study of the Northern-Benghazi Power 3

Optimal Sizing and Allocation of Distributed Energy Resource...

引用

3rd IEEE international Maghreb Meeting of the Conference on Sciences and Techniques of Automatic Control and computer Engineering, MI-STA 2023

作者： Farkash, Noureddine Farag Tahir, Ahmed Engineering University of Benghazi Department of Electrical and Electronics Benghazi Libya

ISBN: (纸本)9798350319897

The main target is to use genetic algorithms to calculate the optimal location and size of distributed Generation (DG) units for power distribution systems. The algorithm's flexibility allows engineers, electric utilities, and distribution network operators to efficiently integrate new DG units into their current distribution networks by making changes and modifications to locate the optimal DG unit and estimate its optimum size. The proposed method can also improve voltage profiles and reduce total real and reactive power losses. Tests conducted on a section of the Benghazi distribution system show that the decision-making algorithm has acceptable accuracy and is compared against results from NEPLAN software. With the increasing popularity of distributed energy systems worldwide, this method provides solutions to many power system problems. © 2023 IEEE.

关键词： Genetic algorithms

来源：评论

学校读者我要写书评

暂无评论

Smart Office Solutions: Monitoring and Enhancing Environmental Quality for Optimal Workplace Performance

Smart Office Solutions: Monitoring and Enhancing Environment...

引用

2024 international symposium on networks, computers and Communications, ISNCC 2024

作者： Gatea, Aseel Habeeb Al-Ibadi, Alaa Ali, Musaab Adil Computer engineering University of Basrah Basrah Iraq

ISBN: (纸本)9798350364910

Smart gadgets are a big part of our life since they make a lot of work easier and simpler, which saves us a lot of time and effort. With time, a lot of employees have put in lengthy workdays and a lot of effort at the office. As a result, sleep deprivation and work stress cause psychological and physical tension, which lowers output. In order to streamline work and shorten working hours, a smart office system has to be designed. The smart office system is made up of intelligent sensors that scan office data and connect it to a smartphone so that it may be managed remotely rather than by pressing. It is made up of a number of systems that increase system security, like the ones that regulate the lighting and temperature and who is allowed to enter the office. Traditional workplaces have been turned into extremely intelligent and efficient spaces through the integration of Artificial Intelligence (AI) into smart offices. This study of the literature explores the development, uses, advantages, difficulties, and potential future developments of AI-powered smart workplaces. It offers a thorough analysis of the current state of the field and highlights significant developments at the nexus of artificial intelligence and office environments. © 2024 IEEE.

关键词： Office automation

来源：评论

学校读者我要写书评

暂无评论

Automated Warehouse 5G Infrastructure Modeling Using Variational Autoencoders

Automated Warehouse 5G Infrastructure Modeling Using Variati...

引用

2024 international symposium on networks, computers and Communications, ISNCC 2024

作者： Gulia, Rahul Ganguly, Amlan Kwasinski, Andres Kuhl, Michael E. Rashedi, Ehsan Hochgraf, Clark Rochester Institute of Technology Department of Computer Engineering RochesterNY United States Rochester Institute of Technology Department of Industrial and Systems Engineering RochesterNY United States Rochester Institute of Technology Department of Electrical and Computer Engineering Technology RochesterNY United States

ISBN: (纸本)9798350364910

The next decade is poised for a transformative shift in wireless communication technologies, driven by the increasing demand for data-intensive applications. Innovations in signal processing, network architecture estimation and design, and spectrum utilization will be critical in realizing the potential of 5G and beyond. These advancements will enable the seamless integration of emerging technologies and empower the digital transformation of industries and society. In this paper, we introduce a Variational Autoencoder (VAE) based model for indoor radio propagation modeling within an automated Industry4.0 warehouse or factory floor in the 5G wireless bands. We detail the creation of training data tensors, the architecture of our proposed VAE model, and its training using tensors that capture the impact of various interacting objects within a 5G-enabled smart warehouse infrastructure. The model is trained on multiple warehouse configurations to predict the signal-to-interference-plus-noise ratio (SINR) for both the validation dataset and unknown warehouse configurations. We present reconstruction error heatmaps to demonstrate the accuracy of our model and analyze its performance in different complex warehouse environments. © 2024 IEEE.

关键词： Signal to noise ratio

来源：评论

学校读者我要写书评

暂无评论

METRICS: A Methodology for Evaluating and Testing the Resilience of Industrial Control systems to Cyberattacks

METRICS: A Methodology for Evaluating and Testing the Resili...

引用

28th European symposium on Research in computer Security (ESORICS)

作者： Bader, Lennart Wagner, Eric Henze, Martin Serror, Martin Fraunhofer FKIE Cyber Anal & Def Wachtberg Germany Rhein Westfal TH Aachen Commun & Distributed Syst Aachen Germany Rhein Westfal TH Aachen Secur & Privacy Ind Cooperat Aachen Germany

ISBN: (纸本)9783031542039;9783031542046

The increasing digitalization and interconnectivity of industrial control systems (ICSs) create enormous benefits, such as enhanced productivity and flexibility, but also amplify the impact of cyberattacks. Cybersecurity research thus continuously needs to adapt to new threats while proposing comprehensive security mechanisms for the ICS domain. As a prerequisite, researchers need to understand the resilience of ICSs against cyberattacks by systematically testing new security approaches without interfering with productive systems. Therefore, one possibility for such evaluations is using already available ICS testbeds and datasets. However, the heterogeneity of the industrial landscape poses great challenges to obtaining comparable and transferable results. In this paper, we propose to bridge this gap with METRICS, a methodology for systematic resilience evaluation of ICSs. METRICS complements existing ICS testbeds by enabling the configuration of measurement campaigns for comprehensive resilience evaluations. Therefore, the user specifies individual evaluation scenarios consisting of cyberattacks and countermeasures while facilitating manual and automatic interventions. Moreover, METRICS provides domain-agnostic evaluation capabilities to achieve comparable results, which user-defined domain-specific metrics can complement. We apply the methodology in a use case study with the power grid simulator Wattson, demonstrating its effectiveness in providing valuable insights for security practitioners and researchers.

关键词： Industrial control systems Security evaluations Testbeds Datasets Resilience

来源：评论

学校读者我要写书评

暂无评论

Detection of Human Falls via computer Vision for Elderly Care – An I3D/RNN Approach 1

引用

20th international symposium on distributed Computing and Artificial Intelligence, DCAI 2023

作者： Leal, João Moayyed, Hamed Vale, Zita Porto Portugal GECAD - Research Group on Intelligent Engineering and Computing for Advanced Innovation and Development LASI - Intelligent Systems Associate Laboratory Polytechnic of Porto Porto Portugal

ISBN: (数字)9783031383335

ISBN: (纸本)9783031383328

As the population continues to age, the number of elderly individuals is increasing at an unprecedented rate as the active age group in developed countries continues to shrink. This demographic shift has resulted in a shortage of resources and limitations in the provision of adequate elderly care. Among the many health risks faced by the elderly, falls are a common and serious problem, which can lead to injuries, hospitalizations, and even death. Despite the prevalence of falls, traditional manual detection methods are often unreliable and inadequate. In response to this, this paper proposes an automatic fall detection system based on video cameras that uses a unique combination of an inflated 3D convolutional neural network (I3D) and a recurrent neural network (RNN). This model was evaluated using multiple fall detection datasets, including a newly developed dataset of simulated falls. The results of this study demonstrate that this hybrid model is highly effective at detecting falls and is on par with other state-of-the-art systems, boasting 94% F1 score and 96% recall values. This technology has the potential to significantly improve the quality of life and safety of the elderly population by providing timely and accurate detection of falls. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.

关键词： computer vision

来源：评论

学校读者我要写书评

暂无评论

Strategy optimization of distributed battery energy storage systems for improving dynamic performances of VRB in active distribution networks

引用

international JOURNAL OF ENERGY RESEARCH 2022年第15期46卷 21812-21825页

作者： Lei, Jiazhi Nanjing Univ Sci & Technol Sch Automat Nanjing Peoples R China

This paper proposed an improved genetic algorithm-based operational strategy for vanadium redox flow battery (VRB) energy storage systems (ESSs) in active distribution networks for improving the dynamic performances of batteries. Firstly, the accurate model of VRB considering the influences of external factors, such as temperature, electrolyte flow rate, ion exchange membrane, catalyst, polarization, self-discharge, and leakage current are constructed. By the test of the accurate model, the dynamic performances Phi of VRB consisting of efficiency eta, self-discharge rate lambda, utilization rate psi(u), maximum discharge depth D-oD, and cycle life kappa are reasonably proposed. And then, the mathematical framework for the operational strategy optimization of ESSs was developed considering both the dynamic performances Phi and the external benefits of VRB ESSs. Finally, case studies based on a modified IEEE 123 Node Test Feeder verified the safe and reasonable operational states of battery ESSs with higher efficiencies, utilization rate, cycle life and lower self-discharge rate, and maximum discharge depth. The dynamic performances of battery ESSs are enhanced by about 32.4%.

关键词： strategy optimization battery energy storage system dynamic performances

来源：评论

学校读者我要写书评

暂无评论

Turbo Table: A Semantic-Aware Cache Acceleration System 22

Turbo Table: A Semantic-Aware Cache Acceleration System

引用

22nd IEEE international symposium on Parallel and distributed Processing with Applications, ISPA 2024

作者： Yuan, Hao Wu, Chentao Li, Jie Guo, Minyi Shanghai Jiao Tong University Department of Computer Science and Engineering Shanghai China

ISBN: (纸本)9798331509712

In the current storage disaggregation architecture, the challenge of quickly retrieving data from storage clusters is typically addressed using caching or data pushdown strategies to accelerate data access and reduce data movement. However, most caching systems still manage data at the page granularity level. For compute-intensive applications, such as analytical applications, this coarse data management approach leads to underutilized computational resources and read-write amplification issues. Additionally, insufficient cache utilization results in inefficient data flow. By managing data at the schema granularity level and modifying the data flow path, we alleviate the read-write amplification problem and improve data transfer speeds. Managing data at the schema level also enables semantic awareness, allowing us to proactively analyze semantics for more precise cache management strategies and accurate prefetching, rather than passively waiting for cache request sequences. We also observed that native compute caches waste valuable cache space and complicate the association between original and result data. To address this, we propose a multi-grained caching model to avoid these limitations. Compared to the baseline, Turbo Table reduces computation time by 2.4% to 36.8%. © 2024 IEEE.

关键词： Information management

来源：评论

学校读者我要写书评

暂无评论

Proceedings - 2024 IEEE international Conference on Edge Computing and Communications, EDGE 2024

Proceedings - 2024 IEEE International Conference on Edge Com...

引用

8th IEEE international Conference on Edge Computing and Communications, EDGE 2024

ISBN: (纸本)9798350368499

The proceedings contain 20 papers. The topics discussed include: split DNN inference for exploiting near-edge accelerators;DONNA: distributed optimized neural network allocation on CIM-based heterogeneous accelerators;reliable network performance for edge networks with QoS-aware adaptive routing;the impact of GPU on containerized computer vision applications on edge nodes;cross network layer cognitive service orchestration in edge computing systems;a case for deploying dynamic neural network on edge-cloud continuum environment;data sharing-aware online algorithms for task allocation in edge computing systems;and SlimNet: a lightweight attentive network for speech-music-noise classification and voice activity detection.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Feature Selection Using COA with Modified Feedforward Neural Network for Prediction of Attacks in Cyber-Security 2

Feature Selection Using COA with Modified Feedforward Neural...

引用

2nd international Conference on distributed Computing and Optimization Techniques, ICDCOT 2024

作者： Vallabhaneni, Rohith Nagamani, H.S. Harshitha, P. Sumanth, S. University of the Cumberlands United States Govt. Smt Vhd Central Institute of Home Science Department of Computer Science Bengaluru India Nitte Meenakshi Institute of Technology Department of Aiml Bengaluru India Government College for Women Department of Computer Science and Applictaions Kolar India

ISBN: (纸本)9798350382952

Research on network intrusion detection, prediction, and mitigation systems has been ongoing due to the exponential rise in cyber-attacks in recent times. The prediction of future network invasions is still an open research subject, despite the abundance of intrusion detection systems. Current solutions need feature selection and engineering since they use statistical and/or shallow machine learning techniques to do the job. This study begins with pre-processing the BotNet dataset with feature encoding and scaling. The next step is to use the Coot Optimisation Algorithm (COA) to pick the most significant characteristics. The next step is for the FeedForward Neural Network (FFNN) model to do the classification. ETLBO is used to determine the model's weight appropriately. Using the adaptive weight approach with the Kent chaotic map improves the ideal performance of the standard TLBO. The projected technique aims to circumvent the primary problems with the original TLBO by not using the local optimum and by maintaining an equilibrium between the search methods. When compared to current methods, the experimental analysis reveals that the suggested model attained a precision of 96.76% and an accuracy of 97.56%. © 2024 IEEE.

关键词： Feedforward neural networks

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：