检索结果-内蒙古大学图书馆

International Conference on High Performance Computing & Simulation (HPCS)

作者： Oyku Melikoglu Oguz Ergin Behzad Salami Julian Pavon Osman Unsal Adrian Cristal Department of Computer Engineering TOBB University of Economics and Technology Ankara Turkey Computer Architecture for Parallel Paradigms Group Barcelona Supercomputing Center Barcelona Spain

ISBN: (数字)9781728144849

ISBN: (纸本)9781728144856

This paper presents a deeply pipelined and massively parallel Binary Search Tree (BST) accelerator for Field Programmable Gate Arrays (FPGAs). Our design relies on the extremely parallel on-chip memory, or Block RAMs (BRAMs) architecture of FPGAs. To achieve significant throughput for the search operation on BST, we present several novel mechanisms including tree duplication as well as horizontal, duplicated, and hybrid (horizontal-vertical) tree partitioning. Also, we present efficient techniques to decrease the stalling rates that can occur during the parallel tree search. By combining these techniques and implementations on Xilinx Virtex-7 VC709 platform, we achieve up to 8X throughput improvement gain in comparison to the baseline implementation, i.e., a fully-pipelined FPGA-based accelerator.

关键词： Registers Throughput Field programmable gate arrays Vegetation Pipeline processing Binary search trees Random access memory

来源：评论

学校读者我要写书评

暂无评论

ml-experiment: A Python framework for reproducible data science

引用

Journal of Physics: Conference Series 2020年第1期1603卷

作者： Antonio Molner Domenech Alberto Guillén Computer Architecture and Technology Department University of Granada Spain.

Nowadays, data science projects are usually developed in an unstructured way, which makes it difficult to reproduce. It is also hard to move from an experimental environment to production. Operational workflows such as containerization, continuous deployment, and cloud orchestration allow data science researchers to move a pipeline from a local environment to the cloud. Being aware of the difficulties of setting those workflows up, this paper presents a framework to ease experiment tracking and operationalizing machine learning by combining existent and well-supported technologies. These technologies include Docker, Mlflow, Ray, among others. The framework provides an opinionated workflow to design and execute experiments either on a local environment or the cloud. ml-experiment includes: an automatic tracking system for the most famous machine learning libraries: Tensorflow, Keras, Fastai, Xgboost and Lightgdm, first-class support for distributed training and hyperparameter optimization, and a Command Line Interface (CLI) for packaging and running projects inside containers.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Self-Adaptive Address Mapping Mechanism for Access Pattern Awareness on DRAM

Self-Adaptive Address Mapping Mechanism for Access Pattern A...

引用

IEEE International Conference on Big Data and Cloud Computing (BdCloud)

作者： Chundian Li Mingzhe Zhang Zhiwei Xu Xianhe Sun SKL of Computer Architecture Institute of Computing Technology Beijing China Department of Computer Science Illinois Institute of Technology Chicago USA University of Chinese Academy of Sciences Beijing China

ISBN: (数字)9781728143286

ISBN: (纸本)9781728143293

As DRAM is a considerably slow storage compared to CPU, the long access latency becomes a serious issue and affects the whole execution if fetching data is on the critical path. It is benefical if the data layout on DRAM, which is decided by address mapping, can serve data accesses with either great locality or bank-level parallelism. However for some cases, there exists a huge mismatch between access patterns and data layout of applications, which introduces the difficulty in obtaining locality or parallelism and current general address mapping cannot resolve it well. In an effort to overcome this challenge, we present an self-adaptive address mapping mechanism in memory controller to be aware of different access patterns on DRAM without any prior knowledge of applications. Moreover, there is nearly no modification on softwares, including applications, libraries and OS. We take several versions of matrix multiplication as an early verification, since their patterns are regular and simple to control. Impressively, the results reveal that memory performance of naive and tiling versions is improved up to 3.4× and 2.7× at most, 2.1× and 1.7× on average respectively. The whole execution time is reduced by up to 24% and 8% averagely. Even for highly-optimized implementations, execution time is decreased by 7% and memory performance speeds up to 1.6× on average.

关键词： Random access memory Layout Parallel processing Organizations Libraries Programming Optimization

来源：评论

学校读者我要写书评

暂无评论

FTT-NAS: Discovering fault-tolerant convolutional neural architecture

arXiv

引用

arXiv 2020年

作者： Ning, Xuefei Ge, Guangjun Li, Wenshuo Zhu, Zhenhua Zheng, Yin Chen, Xiaoming Gao, Zhen Wang, Yu Yang, Huazhong Department of Electronic Engineering Tsinghua University China Weixin Group Tencent China State Key Laboratory of Computer Architecture Institute of Computing Technology Chinese Academy of Sciences China School of Electrical and Information Engineering Tianjin University China

With the fast evolvement of embedded deep-learning computing systems, applications powered by deep learning are moving from the cloud to the edge. When deploying neural networks (NNs) onto the devices under complex environments, there are various types of possible faults: soft errors caused by cosmic radiation and radioactive impurities, voltage instability, aging, temperature variations, malicious attackers, etc. Thus the safety risk of deploying NNs is now drawing much attention. In this paper, after the analysis of the possible faults in various types of NN accelerators, we formalize and implement various fault models from the algorithmic perspective. We propose Fault-Tolerant Neural architecture Search (FT-NAS) to automatically discover convolutional neural network (CNN) architectures that are reliable to various faults in nowadays devices. Then we incorporate fault-tolerant training (FTT) in the search process to achieve better results, which is referred to as FTT-NAS. Experiments on CIFAR-10 show that the discovered architectures outperform other manually designed baseline architectures significantly, with comparable or fewer floating-point operations (FLOPs) and parameters. Specifically, with the same fault settings, F-FTT-Net discovered under the feature fault model achieves an accuracy of 86.2% (VS. 68.1% achieved by MobileNet-V2), and W-FTT-Net discovered under the weight fault model achieves an accuracy of 69.6% (VS. 60.8% achieved by ResNet-18). By inspecting the discovered architectures, we find that the operation primitives, the weight quantization range, the capacity of the model, and the connection pattern have influences on the fault resilience capability of NN models. Copyright © 2020, The Authors. All rights reserved.

关键词： Deep learning

来源：评论

学校读者我要写书评

暂无评论

MV-Sketch: A Fast and Compact Invertible Sketch for Heavy Flow Detection in Network Data Streams

MV-Sketch: A Fast and Compact Invertible Sketch for Heavy Fl...

引用

IEEE Conference on computer Communications

作者： Lu Tang Qun Huang Patrick P. C. Lee Department of Computer Science and Engineering The Chinese University of Hong Kong State Key Lab of Computer Architecture Institute of Computing Technology Chinese Academy of Sciences

ISBN: (纸本)9781728105161

Fast detection of heavy flows (e.g., heavy hitters and heavy changers) in massive network traffic is challenging due to the stringent requirements of fast packet processing and limited resource availability. Invertible sketches are summary data structures that can recover heavy flows with small memory footprints and bounded errors, yet existing invertible sketches incur high memory access overhead that leads to performance degradation. We present MV-Sketch, a fast and compact invertible sketch that supports heavy flow detection with small and static memory allocation. MV-Sketch tracks candidate heavy flows inside the sketch data structure via the idea of majority voting, such that it incurs small memory access overhead in both update and query operations, while achieving high detection accuracy. We present theoretical analysis on the memory usage, performance, and accuracy of MV-Sketch. Trace-driven evaluation shows that MVSketch achieves higher accuracy than existing invertible sketches, with up to 3.38× throughput gain. We also show how to boost the performance of MV-Sketch with SIMD instructions.

关键词： Sketches Reversibility Heavy Packet Processing traffic network degradation of performance Fasting

来源：评论

学校读者我要写书评

暂无评论

Big mobility data reveals hyperlocal air pollution exposure disparities in the Bronx, New York

Nature Cities

引用

Nature Cities 2024年第8期1卷 512-521页

作者： Iacopo Testi An Wang Sanjana Paul Simone Mora Fábio Duarte Paolo Santi Carlo Ratti Erica Walker Marguerite Nyhan Senseable City Lab Department of Urban Studies and Planning Massachusetts Institute of Technology Cambridge MA USA Department of Civil and Environmental Engineering The Hong Kong Polytechnic University Hong Kong SAR China Department of Computer Science Norwegian University of Science and Technology Trondheim Norway Istituto di Informatica e Telematica Consiglio Nazionale delle Ricerche Pisa Italy ABC Department Politecnico di Milano Milan Italy School of Public Health Brown University Providence RI USA School of Engineering and Architecture University College Cork Cork Ireland

Air pollution disproportionately affects socially disadvantaged populations. Our study bridges the existing gap in quantifying mobility-based exposure and its associated disparity issues. We combined the granular mobility of over 500,000 unique anonymized users daily and hyperlocal air pollution data in 100 × 100-m grid cells to quantify disparities in particulate matter exposure in a racially diverse and dense urban area of New York City. Our approach advances the study of exposure and its disparity from individualized exposure tracking to a population-representative scale. We observed apparently different spatial patterns between personal exposure and exposure disparities, noting that people from Hispanic-majority and low-income neighborhoods were those most severely and disproportionately exposed to fine particulate matter (PM2.5) pollution. We reveal that race and ethnicity are much stronger indicators of exposure disparity than income. Our study further demonstrates that within-group variation contributes a major portion to exposure disparities, suggesting more granular mitigation plans are needed to target high-exposure individuals from socially disadvantaged groups in addition to generic air quality improvement.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Corrigendum to ‘Research on Lightweight Anomaly Detection of Multimedia Traffic in Edge Computing’ computer & Security, 111(2021) 102463

引用

computers & Security 2023年 125卷

作者： Xu Zhao Guangqiu Huang Jin Jiang Ling Gao Maozhen Li School of Management Xi'an University of Architecture & Technology Xi'an 710055 China School of Computer Science Xi'an Polytechnic University Xi'an 710048 China Department of Electronic and Electrical Engineering Brunel University London Uxbridge UB8 3PH UK School of Humanities Xi'an Polytechnic University Xi'an 710048 China

来源：评论

学校读者我要写书评

暂无评论

Public perceptions of responsible AI in local government: A multi-country study using the theory of planned behaviour

引用

Government Information Quarterly 2025年第3期42卷

作者： Anne David Tan Yigitcanlar Kevin Desouza Karen Mossberger Pauline Hope Cheong Juan Corchado Prithvi Bhat Beeramoole Alexander Paz School of Architecture and Built Environment Queensland University of Technology 2 George Street Brisbane QLD 4000 Australia School of Management Queensland University of Technology 2 George Street Brisbane QLD 4000 Australia School of Public Affairs Arizona State University Phoenix AZ USA School of Human Communication Arizona State University Phoenix AZ USA Department of Computer Science and Automation University of Salamanca 37007 Salamanca Spain School of Civil and Environmental Engineering Queensland University of Technology 2 George Street Brisbane QLD 4000 Australia

The rise of Artificial Intelligence (AI) has significant implications for local government service delivery, offering considerable advantages alongside inherent risks that warrant careful management. While responsible AI has become a focal point in academic and policy discussions, public perceptions remain marginal in these debates. This paper explores how behavioural factors along with perceived risk, local government AI policy awareness and policy expectations influence public intentions to support local government responsible AI practices. Using the Theory of Planned Behaviour, this study examines a multi-factor survey through Confirmatory Factor Analysis, followed by Structural Equation Modelling to assess relationships between key factors. A survey questionnaire, conducted with participants from Australia, the United States, and Spain via the Prolific platform, reveals key insights: (a) perceived risk exhibit a stronger influence than other factors; (b) policy awareness plays a critical role in shaping public intention towards support for responsible practices; (c) greater AI policy awareness correlates with more realistic expectations of local government AI policies; (d) social influence lacks a notable impact in this context. These findings provide valuable guidance for urban policymakers in crafting AI strategies that promote responsible AI implementation within local government services.

关键词： Responsible AI Urban AI Local government policy Theory of planned behaviour Public intention Policy awareness

来源：评论

学校读者我要写书评

暂无评论

Bringing VFX to the Table 20

Bringing VFX to the Table

引用

ACM SIGGRAPH 2020 Panels - International Conference on computer Graphics and Interactive Techniques, SIGGRAPH 2020

作者： Hanlin, Heath Holmes, Shaina Sayed, Raqi Teich, Sunny Computer Art and Animation Department of Transmedia College of Visual and Performing Arts Syracuse University SyracuseNY United States Vfx and Post Production Department of Television Radio Film S.I. Newhouse School of Public Communications Syracuse University SyracuseNY United States Master of Design Technology Faculty of Architecture and Design School of Design Innovation Victoria University of Wellington Wellington New Zealand Web and Information Design School of Design Innovation Victoria University of Wellington Wellington New Zealand

ISBN: (纸本)9781450379694

This panel brings together academics and practitioners to discuss the structural problems facing visual effects within the film and television industry today. This discussion is organized around four sub-Topics: 1) labor solutions and what unionization would mean, 2) perceptions of VFX artists and their contributions within the industry by filmmakers and artists themselves, as well as the audience, 3) issues of inclusivity along gender and racial lines, and their effects on work/life balances, and 4) VFX as equal creative partners rather than isolated, temporary, highly-skilled specialists. This panel is a continuation of work begun by the faculty and practitioner organizers, Heath Hanlin (Syracuse University, VPA) and Shaina Holmes (Syracuse University, Newhouse) on studying structural challenges and obstacles to entry for students hoping to enter the VFX industry, in particular women, and artists of color. An earlier iteration of this discussion was held at the 2020 Sundance Film Festival. The hope is that the technical focus and high attendance by industry practitioners at Siggraph, adds practitioner perspectives and educator feedback to the sub-Topics listed above. Simply put, today's film and television industry rely heavily on visual effects, both to advance the story or as a means to hide imperfections to keep the audience engaged without technical distractions. Traditional conceptions of VFX artistry as a production step that comes after primary cinematography is simply outdated. Unfortunately, the economic and work-flow structure of the industry accepts this new reality while at the same time attempting to compensate, credit, and organize VFX artists within the outdated framework. Often this aspect of the conversation is moored in debates surrounding the potentiality of AI to either positively impact the industry (no more underpaid routinized tasks for artists) or negatively (crafts persons and artists can be replaced by algorithms). Rather, a more fruitful

关键词： Iterative methods

来源：评论

学校读者我要写书评

暂无评论

Computational prediction of the stability of tensegrity structures 7

Computational prediction of the stability of tensegrity stru...

引用

7th International Conference on Computational Methods in Structural Dynamics and Earthquake Engineering, COMPDYN 2019

作者： Bieniek, Z. Mascolo, I. Amendola, A. Micheletti, A. Luciano, R. Fraternali, F. Faculty of Civil and Environmental Engineering and Architecture Rzeszów University of Technology Rzeszów Poland Department of Civil Engineering University of Salerno Italy Department of Civil Engineering and Computer Science University of Rome Tor Vergata Italy Department of Engineering University of Naples "Parthenope" Naples Italy

ISBN: (纸本)9786188284470

The lightweight nature of tensegrity structures calls for the formulation of computational tools that are able to analyze the stability problem of such structures, both in statics and dynamics. The present work analyzes the concepts of prestress-stability and superstability with reference to a special class of tensegrity structures recently appeared in the literature, named class theta tensegrity prisms. Such structures are formed by two triangular prisms superimposed one over the other, and two different sets of strings connecting the two prisms. The given results can be easily generalized to general tensegrity structures and can be usefully incorporated in form-finding and topology optimization codes. © 2019 The authors.

关键词： Stability

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：