检索结果-内蒙古大学图书馆

32nd ACM Symposium on Parallelism in Algorithms and architectures, SPAA 2020

作者： Zhang, Xiaoyang Xiao, Junmin Tan, Guangming State Key Laboratory of Computer Architecture Institute of Computing Technology Chinese Academy of Sciences Beijing China University of Chinese Academy of Science Beijing China

ISBN: (纸本)9781450369350

Convolution is the most time-consuming part in the computation of convolutional neural networks (CNNs). Due to the complex data dependency and the increase in the amount of model samples, the convolution suffers from high overhead on data movement. This work provides comprehensive analysis and methodologies to minimize the communication for the convolutions in CNNs. With an in-depth analysis on the I/O complexity theory under the red-blue pebble game model, we develop a general communication lower bound theory for a composite algorithm which consists of several different sub-computations. Based on the proposed theory, we establish the data movement lower bound results for three main convolution algorithms in CNNs, which are the direct convolution, the image2col method and Winograd algorithm. Furthermore, derived from I/O lower bound results, we design the near communication-optimal strategies respectively for the three main convolution algorithms by fully exploiting the data reuse. The deep analysis demonstrates that our designs are able to nearly reach the minimum communication in a two-level memory hierarchy. © 2020 Owner/Author.

关键词： Convolution

来源：评论

学校读者我要写书评

暂无评论

Exploiting software architecture to support requirements satisfaction testing

Exploiting software architecture to support requirements sat...

引用

19th ACM SIGSOFT Symposium on Foundations of Software Engineering, SIGSOFT/FSE'11

作者： Clements, Paul Escalona, María J. Inverardi, Paola Malavolta, Ivano Marchetti, Eda Software Engineering Institute Carnegie Mellon University United States Department of Languages and Computer Systems University of Sevilla Spain Department of Computer Science University of Aquila Italy Software Engineering Laboratory ISTI-CNR Pisa Italy

ISBN: (纸本)9781450304436

Currently, software testing is mainly carried on independently from software architecture-related information. Some approaches propose to perform integration and regression testing with respect to software architecture descriptions, but less attention has been paid to analysing software architecture in order to develop a less costly and time-consuming test plan that covers the requirements of the system of interest. If on one side, it is well known that providing an effective test plan is crucial to software quality, on the other side software testing is extremely difficult because it stems from the complexity of current software systems. In this paper we (i) elaborate on how a new architectural analysis can help in producing better test plans and (ii) identify a set of corresponding research challenges. We believe that considering and analysing architectural information for requirements satisfaction testing purposes will provide substantial benefits in terms of test plan specification process, test plan effectiveness, and test cases understandability. © 2011 ACM.

关键词： Software architecture

来源：评论

学校读者我要写书评

暂无评论

Fault-tolerant middleware based on multistream pipeline for private storage services

Fault-tolerant middleware based on multistream pipeline for ...

引用

7th International Conference for Internet Technology and Secured Transactions, ICITST 2012

作者： Gonzalez, J.L. Sosa-Sosa, Victor Bergua, Borja Sanchez, Luis Miguel Carretero, Jesus Cd. Valles Institute of Technology Information Technology Laboratory Center of Research CINVESTAV Ciudad Victoria Mexico Computer Architecture and Technology Area Computer Science Department Universidad Carlos III Madrid Spain

ISBN: (纸本)9781908320087

This paper presents a fault-tolerant middleware for private storage services based on a client-server model. A client-side API split files into redundant chunks, which are encoded/decoded, anonymized, and distributed to different storage locations in parallel by using multistream pipeline. A chunk placement engine on the server-side API receives/delivers chunks from/to the client side API. In this model, the user preserves her files in-house and achieves data availability anytime from anywhere even in site failure scenarios while the organizations preserves the control of their storage resources. The experimental evaluation reveals the middleware can compensate the overhead produced on the client side by taking advantage of multiple cores commonly found in user's devices. The users observe that the middleware produces similar performance to the private and public file hosting systems tested in the evaluation. The organizations observe a significant reduction of work in their servers. In site disaster scenarios, the middleware yields even better performance than fault-tolerant online distributed solutions tested in the evaluation. © 2012 Infonomics Society.

关键词： Middleware

来源：评论

学校读者我要写书评

暂无评论

Small Object Detection using Multi-scale Feature Fusion and Attention 41

Small Object Detection using Multi-scale Feature Fusion and ...

引用

第41届中国控制会议

作者： Baokai Liu Shiqiang Du Jiacheng Li Jianhua Wang Wenjie Liu Key Laboratory of Chinas Ethnic Languages and Information Technology of Ministry of Education Chinese National Information Technology Research InstituteNorthwest Minzu University College of Mathematics and Computer Science Northwest Minzu University

ISBN: (数字)9789887581536

ISBN: (纸本)9781665482561

In deep learning,object detection has achieved a very large performance ***,due to the few features available for small objects,network structure,sample imbalance and other reasons,the result is unsatisfactory in small object *** order to solve this problem,this paper proposes a method based on the combination of mutil-scale feature fusion and dilated convolution,which uses dilated convolution to expand the receptive field of feature maps at different scales and then extracts the high-level semantic information and low-level semantic information from the backbone *** obtained feature maps of different receptive fields are fused to obtain the final feature map prediction *** addition,we add a series of channel attention and spatial attention mechanisms to the network to better obtain the context information of the object in the *** show that this method can have higher accuracy than the traditional YOLOv3 network in the detection of small *** addition,the size of 640*640 images,we can achieve 31.5% accuracy in the detection of small objects in MS *** with YOLOv5,there are 4 points of improvement.

关键词： Attention Mechanism Dilated Convolution Multi-scale Feature Fusion Channel Attention Spatial Attention

来源：评论

学校读者我要写书评

暂无评论

Multi-view Spectral Clustering Based on Graph Learning 41

Multi-view Spectral Clustering Based on Graph Learning

引用

第41届中国控制会议

作者： Jinmei Song Baokai Liu Kaiwu Zhang Yao Yu Shiqiang Du Key Laboratory of Chinas Ethnic Languages and Information Technology of Ministry of Education Chinese National Information Technology Research InstituteNorthwest Minzu University College of Mathematics and Computer Science Northwest Minzu University

ISBN: (数字)9789887581536

ISBN: (纸本)9781665482561

Multi-view clustering is based on the exploration of complementary information from different *** graphbased approach is a representative method among existing multi-view clustering ***,how to effectively obtain complementary information in multi-view data and how to perform effective clustering in the face of the low quality of the initially constructed data graph is still a *** address the aforementioned issues,we propose a new multi-view spectral clustering model,namely multi-view spectral clustering based on graph learning(GLSC),which utilizes the relevant knowledge of self-representing attributes to construct similarity graphs that can represent the relationship between data,uses the constrained Laplacian rank method to construct an affinity graph for the similarity graphs,and applies the classical spectral clustering method for *** propose an iterative algorithm to solve the optimization ***,we conduct extensive experiments on four real benchmark datasets,and the experimental results show the superiority of GLSC clustering performance.

关键词： Multi-view Clustering Spectral Clustering Graph Learning Constrained Laplacian Rank

来源：评论

学校读者我要写书评

暂无评论

LACC:a hardware and software co-design accelerator for deep neural networks

引用

High Technology Letters 2021年第1期27卷 62-67页

作者： Yu Yong Zhi Tian Zhou Shengyuan State Key Laboratory of Computer Architecture Institute of Computing TechnologyChinese Academy of SciencesBeijing 100190P.R.China School of Computer Science and Technology University of Chinese Academy of SciencesBeijing 100049P.R.China Cambricon Technologies Ltd Beijing 100190P.R.China

With the increasing of data size and model size,deep neural networks(DNNs)show outstanding performance in many artificial intelligence(AI)*** the big model size makes it a challenge for high-performance and low-power running DNN on processors,such as central processing unit(CPU),graphics processing unit(GPU),and tensor processing unit(TPU).This paper proposes a LOGNN data representation of 8 bits and a hardware and software co-design deep neural network accelerator LACC to meet the *** data representation replaces multiply operations to add and shift operations in running *** accelerator achieves higher efficiency than the state-of-the-art DNN accelerators by domain specific arithmetic computing ***,LACC speeds up the performance per watt by 1.5 times,compared to the state-of-the-art DNN accelerators on average.

关键词： deep neural network(DNN) domain specific accelerator domain specific data type

来源：评论

学校读者我要写书评

暂无评论

FACET: PC-parallel architecture for cost-efficient telemetry processing

FACET: PC-parallel architecture for cost-efficient telemetry...

引用

1998 IEEE Aerospace Conference, AERO 1998

作者： El-Ghazawi, Tarek Chalermwat, Prachya Piamsa-Nga, Punpiti Ozkaya, Armagan Speciale, Nick Wilson, Donald Computer Engineering and Computer Science Florida Institute of Technology MelbourneFL312901-6975 United States Parallel Architecture Laboratory Department of Electrical Engineering and Computer Science George Washington University WashingtonDC20052 United States NASA GSFC GreenbeltMD20771 United States

ISBN: (纸本)0780343115

NASA's Mission to Planet Earth (MTPE) is planning to launch the Earth Observing System (EOS) starting in 1998. The large number of planned remote sensing satellites will bring 500 Gigabytes of information per day. The EOS Data and Information System (EOSDIS) is responsible for ingesting and archiving this data. One important component of the EOSDIS system is the data operation, which involves extracting the packets and reconstructing and archiving the original remotely sensed data products. Due to transmission errors, the way data is sampled from the different sensors encoded, packets typically arrive out of order and perhaps with some of them missing or repeated. Many special hardware solutions have been proposed to solve this real-time problem. In this paper, we demonstrate a commercial off the shelf (COTS) solution. The hardware capitalizes on the progress made in the area of network of workstations (NOW), particularly PC-clusters. The software and algorithm exploit the data characteristics and parallelism in the telemetry stream to make use of load balancing and efficient parallel processing. It is shown that this solution can provide high-performance to cost and programmability. © 1998 IEEE.

关键词： Parallel architectures

来源：评论

学校读者我要写书评

暂无评论

A scalable approach to column-based low-rank matrix approximation

A scalable approach to column-based low-rank matrix approxim...

引用

23rd International Joint Conference on Artificial Intelligence, IJCAI 2013

作者： Pi, Yifan Peng, Haoruo Zhou, Shuchang Zhang, Zhihua Institute for Theoretical Computer Science IIIS Tsinghua University Beijing 100084 China Department of Computer Science and Technology Tsinghua University Beijing 100084 China State Key Laboratory of Computer Architecture Institute of Computing Technology Chinese Academy of Sciences Beijing 100190 China Google Inc. Beijing 100084 China College of Computer Science and Technology Zhejiang University Hangzhou 310027 China

ISBN: (纸本)9781577356332

In this paper, we address the column-based low-rank matrix approximation problem using a novel parallel approach. Our approach is based on the divide-and-combine idea. We first perform column selection on submatrices of an original data matrix in parallel, and then combine the selected columns into the final output. Our approach enjoys a theoretical relative-error upper bound. In addition, our column-based low-rank approximation partitions data in a deterministic way and makes no assumptions about matrix coherence. Compared with other traditional methods, our approach is scalable on largescale matrices. Finally, experiments on both simulated and real world data show that our approach is both efficient and effective.

关键词： Matrix algebra

来源：评论

学校读者我要写书评

暂无评论

TransferTOD: A Generalizable Chinese Multi-Domain Task-Oriented Dialogue System with Transfer Capabilities

TransferTOD: A Generalizable Chinese Multi-Domain Task-Orien...

引用

2024 Conference on Empirical Methods in Natural Language Processing, EMNLP 2024

作者： Zhang, Ming Huang, Caishuang Wu, Yilong Liu, Shichun Zheng, Huiyuan Dong, Yurui Shen, Yujiong Dou, Shihan Zhao, Jun Ye, Junjie Zhang, Qi Gui, Tao Huang, Xuanjing School of Computer Science Fudan University China Shanghai Key Laboratory of Intelligent Information Processing Fudan University China Institute of Modern Languages and Linguistics Fudan University China

ISBN: (纸本)9798891761643

Task-oriented dialogue (TOD) systems aim to efficiently handle task-oriented conversations, including information collection. How to utilize TOD accurately, efficiently and effectively for information collection has always been a critical and challenging task. Recent studies have demonstrated that Large Language Models (LLMs) excel in dialogue, instruction generation, and reasoning, and can significantly enhance the performance of TOD through fine-tuning. However, current datasets primarily cater to user-led systems and are limited to predefined specific scenarios and slots, thereby necessitating improvements in the proactiveness, diversity, and capabilities of TOD. In this study, we present a detailed multi-domain task-oriented data construction process for conversations, and a Chinese dialogue dataset generated based on this process, TransferTOD, which authentically simulates human-computer dialogues in 30 popular life service scenarios. Leveraging this dataset, we trained a model using full-parameter fine-tuning called TransferTOD-7B, showcasing notable abilities in slot filling and questioning. Our work has demonstrated its strong generalization capabilities in various downstream scenarios, significantly enhancing both data utilization efficiency and system performance. The data is released in https://***/KongLongGeFDU/TransferTOD. © 2024 Association for Computational Linguistics.

关键词： Computational linguistics

来源：评论

学校读者我要写书评

暂无评论

ProofInfer: Generating Proof via Iterative Hierarchical Inference

ProofInfer: Generating Proof via Iterative Hierarchical Infe...

引用

2022 Conference on Empirical Methods in Natural Language Processing, EMNLP 2022

作者： Fei, Zichu Zhang, Qi Zhou, Xin Gui, Tao Huang, Xuanjing School of Computer Science Fudan Unviersity China Shanghai Key Laboratory of Intelligent Information Processing Shanghai China Institute of Modern Languages and Linguistics Fudan University Shanghai China

Proof generation focuses on deductive reasoning: given a hypothesis and a set of theories, including some supporting facts and logical rules expressed in natural language, the model generates a proof tree indicating how to deduce the hypothesis from given theories. Current models with state-of-the-art performance employ the stepwise method, linking an individual node to the proof step-by-step. However, these methods actually focus on generating several proof paths rather than a whole tree. To address this problem, we propose ProofInfer, which generates the proof tree via iterative hierarchical inference. At each step, ProofInfer generates the entire layer for proof tree, where all nodes in this layer are generated simultaneously. Since the conventional autoregressive generation architecture cannot simultaneously predict multiple nodes, ProofInfer employs text-to-text paradigm to avoid it. To this end, we propose a divide- and-conquer algorithm to encode the proof tree as the plain text recursively without structure information loss. Experimental results show that ProofInfer significantly outperforms the state-of-the-art (SOTA) models on several widely-used datasets. In addition, ProofInfer still performs well with data-limited, achieving comparable performance to the SOTA models with only 40% of the training data. © 2022 Association for Computational Linguistics.

关键词： Iterative methods

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：