检索结果-内蒙古大学图书馆

Towards optimized tensor code generation for deep learning on sunway many-core processor

Frontiers of computer science 2024年第2期18卷 1-15页

作者： Mingzhen LI Changxi LIU Jianjin LIAO Xuegui ZHENG Hailong YANG Rujun SUN Jun XU Lin GAN Guangwen YANG Zhongzhi LUAN Depei QIAN State Key Laboratory of Software Development Environment Beijing 100191China School of Computer Science and Engineering Beihang UniversityBeijing 100191China National University of Singapore Singapore 119077Singapore State Key Laboratory of Mathematical Engineering and Advanced Computing Wuxi 214000China Science and Technology on Special System Simulation Laboratory Beijing Simulation Center Beijing 100854China Department of Computer Science and Technology Tsinghua UniversityBeijing 100084China

The flourish of deep learning frameworks and hardware platforms has been demanding an efficient compiler that can shield the diversity in both software and hardware in order to provide application *** the existing deep learning compilers,TVM is well known for its efficiency in code generation and optimization across diverse hardware *** the meanwhile,the Sunway many-core processor renders itself as a competitive candidate for its attractive computational power in both scientific computing and deep learning *** paper combines the trends in these two ***,we propose swTVM that extends the original TVM to support ahead-of-time compilation for architecture requiring cross-compilation such as *** addition,we leverage the architecture features during the compilation such as core group for massive parallelism,DMA for high bandwidth memory transfer and local device memory for data locality,in order to generate efficient codes for deep learning workloads on *** experiment results show that the codes generated by swTVM achieve 1.79x improvement of inference latency on average compared to the state-of-the-art deep learning framework on Sunway,across eight representative *** work is the first attempt from the compiler perspective to bridge the gap of deep learning and Sunway processor particularly with productivity and efficiency in *** believe this work will encourage more people to embrace the power of deep learning and Sunwaymany-coreprocessor.

关键词： sunway processor deep learning compiler code generation performance optimization

来源：评论

学校读者我要写书评

暂无评论

Research on performance optimization of virtual data space across WAN

引用

Frontiers of computer science 2024年第6期18卷 167-187页

作者： Jiantong HUO Zhisheng HUO Limin XIAO Zhenxue HE State Key Laboratory of Software Development Environment Beihang UniversityBeijing 100191China School of Computer Science and Engineering Beihang UniversityBeijing 100191China High Performance Computing Center Beihang UniversityBeijing 100191China Hebei Key Laboratory of Agricultural Big Data Hebei Agricultural UniversityBaoding 071001China

For the high-performance computing in a WAN environment,the geographical locations of national supercomputing centers are scattered and the network topology is complex,so it is difficult to form a unified view of *** aggregate the widely dispersed storage resources of national supercomputing centers in China,we have previously proposed a global virtual data space named GVDS in the project of“High Performance Computing Virtual Data Space”,a part of the National key Research and development Program of *** GVDS enables large-scale applications of the high-performance computing to run efficiently across ***,the applications running on the GVDS are often data-intensive,requiring large amounts of data from multiple supercomputing centers across *** this regard,the GVDS suffers from performance bottlenecks in data migration and access across *** solve the above-mentioned problem,this paper proposes a performance optimization framework of GVDS including the multitask-oriented data migration method and the request access-aware IO proxy resource allocation *** a WAN environment,the framework proposed in this paper can make an efficient migration decision based on the amount of migrated data and the number of multiple data sources,guaranteeing lower average migration latency when multiple data migration tasks are running in *** addition,it can ensure that the thread resource of the IO proxy node is fairly allocated among different types of requests(the IO proxy is a module of GVDS),so as to improve the application’s performance across *** experimental results show that the framework can effectively reduce the average data access delay of GVDS while improving the performance of the application greatly.

关键词： storage aggregation across WANs large-scale applications GVDS data migration allocation of IO proxy resource

来源：评论

学校读者我要写书评

暂无评论

Effective Training of PINNs by Combining CMA-ES with Gradient Descent 13

Effective Training of PINNs by Combining CMA-ES with Gradien...

引用

13th IEEE Congress on Evolutionary Computation, CEC 2024

作者： Liu, Lin Yuan, Yuan School of Computer Science and Engineering Beihang University Beijing100191 China State Key Laboratory of Software Development Environment Zhongguancun Laboratory Beijing100191 China

ISBN: (纸本)9798350308365

Physics-Informed Neural Networks (PINNs) have recently received increasing attention, however, optimizing the loss function of PINNs is notoriously difficult, where the landscape of the loss function is often highly non-convex and rugged. Local optimization methods based on gradient information can converge quickly but are prone to being trapped in local minima for training PINNs. Evolutionary algorithms (EAs) are well known for the global search ability, which can help escape from local minima. It has been reported in the literature that EAs show some advantages over gradient-based methods in training PINNs. Inspired by the Memetic Algorithm, we combine global-search based EAs and local-search based batch gradient descent in order to make the best of both word. In addition, since the PINN loss function is composed of multiple terms, balancing these terms is also a challenging issue. Therefore, we also attempt to combine EAs with multiple-gradient descent algorithm for multi-objective optimization. Our experiments provide strong evidence for the superiority of the above algorithms. © 2024 IEEE.

关键词： Optimization algorithms

来源：评论

学校读者我要写书评

暂无评论

Light field depth estimation:A comprehensive survey from principles to future

引用

High-Confidence Computing 2024年第1期4卷 92-107页

作者： Tun Wang Hao Sheng Rongshan Chen Da Yang Zhenglong Cui Sizhe Wang Ruixuan Cong Mingyuan Zhao State Key Laboratory of Virtual Reality Technology and Systems School of Computer Science and EngineeringBeihang UniversityBeijing 100191China Zhongfa Aviation Institute Beihang UniversityHangzhou 311115China Faculty of Applied Sciences Macao Polytechnic UniversityMacao 999078China State Key Laboratory of Software Development Environment School of Computer Science and EngineeringBeihang UniversityBeijing 100191China

Light Field(LF)depth estimation is an important research direction in the area of computer vision and computational photography,which aims to infer the depth information of different objects in threedimensional scenes by capturing LF *** this new era of significance,this article introduces a survey of the key concepts,methods,novel applications,and future trends in this *** summarize the LF depth estimation methods,which are usually based on the interaction of radiance from rays in all directions of the LF data,such as epipolar-plane,multi-view geometry,focal stack,and deep *** analyze the many challenges facing each of these approaches,including complex algorithms,large amounts of computation,and speed *** addition,this survey summarizes most of the currently available methods,conducts some comparative experiments,discusses the results,and investigates the novel directions in LF depth estimation.

关键词： Light field Depth estimation Deep learning Sub-aperture image Epipolar-plane image

来源：评论

学校读者我要写书评

暂无评论

Adaptive Cost Aggregation in Iterative Depth Estimation for Efficient Multi-view Stereo 12th

Adaptive Cost Aggregation in Iterative Depth Estimation for...

引用

12th International Conference on Image and Graphics, ICIG 2023

作者： Wang, Xiang Bai, Xiao Wang, Chen School of Computer Science and Engineering State Key Laboratory of Software Development Environment Jiangxi Research Institute Beihang University Beijing China

ISBN: (纸本)9783031463075

The deep multi-view stereo (MVS) approaches generally construct 3D cost volumes to regularize and regress the depth map. These methods are limited with high-resolution outputs since the memory and time costs grow cubically as the volume resolution increases. In this paper, we presented an multi-stage iterative depth map estimation method for MVS. In our network, the cost volume is iteratively processed by lightweight 2D convolution based GRU modules, and the multi-stage coarse-to-fine structure is adopted to speed up the depth estimation process. To further improve the 3D reconstruction quality, we make improvements from two different perspectives of adaptive cost aggregation: a view-adaptive weighting module is proposed to account for the occlusion problem in cost volume fusion, and a spatial-adaptive deformable geometric feature encoding module is introduced to the cost volume feature encoding before feeding into GRUs for stronger modeling capability. Experiments on the DTU dataset demonstrated the effectiveness of the proposed network in accuracy with remarkable efficiency performance. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023.

关键词： Iterative methods

来源：评论

学校读者我要写书评

暂无评论

Distributed Truss Computation in Dynamic Graphs

引用

Tsinghua science and Technology 2023年第5期28卷 873-887页

作者： Ziwei Mo Qi Luo Dongxiao Yu Hao Sheng Jiguo Yu Xiuzhen Cheng School of Computer Science and Technology Shandong UniversityQingdao 266200China State Key Laboratory of Software Development Environment School of Computer Science and Engineering and the Beijing Advanced Innovation Center for Big Data and Brain ComputingBeihang UniversityBeijing 100191China Big Data Institute Qilu University of Technology(Shandong Academy of Sciences)Jinan 250353China

Large-scale graphs usually exhibit global sparsity with local cohesiveness,and mining the representative cohesive subgraphs is a fundamental problem in graph *** k-truss is one of the most commonly studied cohesive subgraphs,in which each edge is formed in at least k 2 triangles.A critical issue in mining a k-truss lies in the computation of the trussness of each edge,which is the maximum value of k that an edge can be in a *** works mostly focus on truss computation in static graphs by sequential ***,the graphs are constantly changing dynamically in the real *** study distributed truss computation in dynamic graphs in this *** particular,we compute the trussness of edges based on the local nature of the k-truss in a synchronized node-centric distributed *** decomposing the trussness of edges by relying only on local topological information is possible with the proposed distributed decomposition ***,the distributed maintenance algorithm only needs to update a small amount of dynamic information to complete the *** experiments have been conducted to show the scalability and efficiency of the proposed algorithm.

关键词： distributed algorithm dynamic graph graph mining cohesive subgraph k-truss

来源：评论

学校读者我要写书评

暂无评论

FL-GNN: Efficient Fusion of Fuzzy Neural Network and Graph Neural Network 27

FL-GNN: Efficient Fusion of Fuzzy Neural Network and Graph N...

引用

27th European Conference on Artificial Intelligence, ECAI 2024

作者： Du, Boyu Zhou, Jingya Liu, Ling She, Xiaolong School of Computer Science and Technology Soochow University China Engineering Lab.of Big Data and Intelligence of Jiangsu Province Soochow University China State Key Lab.for Novel Software Technology Nanjing University China School of Computer Science Georgia Institute of Technology United States

ISBN: (纸本)9781643685489

Graph Neural Networks (GNNs) have become an important graph feature learning paradigm that is extensively applied to graph inference ***, GNNs still have limitations in some aspects such as representation capability, interpretability, *** this end, this paper introduces a novel hybrid Fuzzy-Logic Graph Neural Network (FL-GNN) that synergistically combines Fuzzy Neural Network (FNN) and GNN to effectively capture and aggregate local information flows within the ***-GNN exhibits three distinctive ***, we incorporate a specialized structure fuzzy rule to enhance FL-GNN's graph inference capability, surpassing representative GNN ***, we augment the interpretability of FL-GNN by integrating analytical exploration methods from two perspectives: Fuzzy Inference System and Message Passing Algorithm (MPA).Lastly, we refine the structure of FL-GNN based on MPA to optimize its calculation complexity and consequently improve learning *** results show that FL-GNN outperforms existing representative graph neural networks for graph inference tasks. © 2024 The Authors.

关键词： Graph neural networks

来源：评论

学校读者我要写书评

暂无评论

A study on the impact of pre-trained model on Just-In-Time defect prediction 23

A study on the impact of pre-trained model on Just-In-Time d...

引用

23rd IEEE International Conference on software Quality, Reliability, and Security, QRS 2023

作者： Guo, Yuxiang Gao, Xiaopeng Zhang, Zhenyu Chan, W.K. Jiang, Bo Beihang University State Key Laboratory of Software Development Environment School of Computer Science and Engineering Beijing China State Key Laboratory of Computer Science Institute of Software Chinese Academy of Sciences Beijing China City University of Hong Kong Department of Computer Science Hong Kong

ISBN: (纸本)9798350319583

Previous researchers conducting Just-In-Time (JIT) defect prediction tasks have primarily focused on the performance of individual pre-trained models, without exploring the relationship between different pre-trained models as backbones. In this study, we build six models: RoBERTaJIT, CodeBERTJIT, BARTJIT, PLBARTJIT, GPT2JIT, and CodeGPTJIT, each with a distinct pre-trained model as its backbone. We systematically explore the differences and connections between these models. Specifically, we investigate the performance of the models when using Commit code and Commit message as inputs, as well as the relationship between training efficiency and model distribution among these six models. Additionally, we conduct an ablation experiment to explore the sensitivity of each model to inputs. Furthermore, we investigate how the models perform in zero-shot and few-shot scenarios. Our findings indicate that each model based on different backbones shows improvements, and when the backbone's pre-training model is similar, the training resources that need to be consumed are closer. We also observe that Commit code plays a significant role in defect detection, and different pre-trained models demonstrate better defect detection ability with a balanced dataset under few-shot scenarios. These results provide new insights for optimizing JIT defect prediction tasks using pre-trained models and highlight the factors that require more attention when constructing such models. Additionally, CodeGPTJIT and GPT2JIT achieved better performance than DeepJIT and CC2Vec on the two datasets respectively under 2000 training samples. These findings emphasize the effectiveness of transformer-based pre-trained models in JIT defect prediction tasks, especially in scenarios with limited training data. © 2023 IEEE.

关键词： Forecasting

来源：评论

学校读者我要写书评

暂无评论

Aster: Encoding Data Augmentation Relations into Seed Test Suites for Robustness Assessment and Fuzzing of Data-Augmented Deep Learning Models 23

Aster: Encoding Data Augmentation Relations into Seed Test S...

引用

23rd IEEE International Conference on software Quality, Reliability, and Security, QRS 2023

作者： Wang, Haipeng Wei, Zhengyuan Zhou, Qilin Jiang, Bo Chan, W.K. City University of Hong Kong Hong Kong Beihang University State Key Laboratory of Software Development Environment School of Computer Science and Engineering Beijing China

ISBN: (纸本)9798350319583

Data-augmented deep learning models are widely used in real-world applications. However, many state-of the-art loss-based or coverage-based fuzzing techniques fail to produce fuzzing samples for them from many seeds. This paper proposes Aster, a novel technique to address this problem to enhance their fuzzing effectiveness for deep learning models trained with multi-sample data augmentation methods. Aster formulates a novel reachability-based strategy to encode the insights of every seed's direct and indirect data augmentation relation instances into the replacement seed of that seed systematically. Our experiment shows that Aster is highly effective. On average, loss-based and coverage-based fuzzing techniques can generate 166% and 110% more fuzzing samples and reduce 31% and 22% unsuccessful seeds, respectively, after adopting the replacement seeds generated by Aster to replace their original seeds. Their improved models also become up to 55% and 40% on average more robust against FGSM and PGD attacks in the experiment. © 2023 IEEE.

关键词： Encoding (symbols)

来源：评论

学校读者我要写书评

暂无评论

Light field super-resolution using complementary-view feature attention

引用

Computational Visual Media 2023年第4期9卷 843-858页

作者： Wei Zhang Wei Ke Da Yang Hao Sheng Zhang Xiong Faculty of Applied Sciences Macao Polytechnic UniversityMacao SAR 999078China State Key Laboratory of Software Development Environment School of Computer Science and EngineeringBeihang UniversityBeijing 100191Chinaand Beihang Hangzhou Innovation Institute YuhangXixi Octagon CityYuhang DistrictHangzhou 310023China

Light field(LF)cameras record multiple perspectives by a sparse sampling of real scenes,and these perspectives provide complementary *** information is beneficial to LF super-resolution(LFSR).Compared with traditional single-image super-resolution,LF can exploit parallax structure and perspective correlation among different LF ***,the performance of existing methods are limited as they fail to deeply explore the complementary information across LF *** this paper,we propose a novel network,called the light field complementary-view feature attention network(LF-CFANet),to improve LFSR by dynamically learning the complementary information in LF ***,we design a residual complementary-view spatial and channel attention module(RCSCAM)to effectively interact with complementary information between complementary ***,RCSCAM captures the relationships between different channels,and it is able to generate informative features for reconstructing LF images while ignoring redundant ***,a maximum-difference information supplementary branch(MDISB)is used to supplement information from the maximum-difference angular positions based on the geometric structure of LF *** branch also can guide the process of *** results on both synthetic and real-world datasets demonstrate the superiority of our *** proposed LF-CFANet has a more advanced reconstruction performance that displays faithful details with higher SR accuracy than state-of-the-art methods.

关键词： light field(LF) super-resolution(SR) attention

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：