With the practical experience of constru-ction bidding documentation,for example,in view of the large infrastructure project construction in colleges and universities bidding documents for the main body,the constructi...With the practical experience of constru-ction bidding documentation,for example,in view of the large infrastructure project construction in colleges and universities bidding documents for the main body,the construction technology,qualification,performance requirements,bill of quantities,the terms of the contract set aspects were discussed,and put forward practical measures and methods,for similar project construction bidding document preparation to provide certain reference.展开更多
Investing in large transport projects affects the (potential) economic development of metropolitan areas. Yet, very little critical research has been performed to understand how to assess these effects. The relationsh...Investing in large transport projects affects the (potential) economic development of metropolitan areas. Yet, very little critical research has been performed to understand how to assess these effects. The relationship between infrastructure investments and regional economic development is complex and indirect, and many theoretical and methodological difficulties remain. On the one hand, the assumption that investing in infrastructure is important to sustain economic growth is sometimes doubted. On the other hand, it is argued that investments in infrastructure enhance the accessibility of urban regions and that in the slipstream of such investments, social problems in urban regions can be tackled as well. Despite these contrasting views, there is at least a consensus that transport infrastructure development depends on economic development and vice versa. Yet, in many cases, the method of assessing economic impacts highly affects the results. Therefore, this paper focuses on a critical reflection of methods for estimating economic effects of infrastructure investments. A critical evaluation is made based on Indonesian and Japanese cases. After conducting in-depth desk research on both cases, we found that the broader effects on affected group of people tend to be overlooked due to the problems of time and space dimensions, the chain reaction of effects, and inappropriate data practices. The assessment on the appraisal processes tends to overlook the broader economic implication due to narrow focus and the concept of efficiency of economic theory.展开更多
The frequent occurrence of extreme weather events has rendered numerous landslides to a global natural disaster issue.It is crucial to rapidly and accurately determine the boundaries of landslides for geohazards evalu...The frequent occurrence of extreme weather events has rendered numerous landslides to a global natural disaster issue.It is crucial to rapidly and accurately determine the boundaries of landslides for geohazards evaluation and emergency response.Therefore,the Skip Connection DeepLab neural network(SCDnn),a deep learning model based on 770 optical remote sensing images of landslide,is proposed to improve the accuracy of landslide boundary detection.The SCDnn model is optimized for the over-segmentation issue which occurs in conventional deep learning models when there is a significant degree of similarity between topographical geomorphic features.SCDnn exhibits notable improvements in landslide feature extraction and semantic segmentation by combining an enhanced Atrous Spatial Pyramid Convolutional Block(ASPC)with a coding structure that reduces model complexity.The experimental results demonstrate that SCDnn can identify landslide boundaries in 119 images with MIoU values between 0.8and 0.9;while 52 images with MIoU values exceeding 0.9,which exceeds the identification accuracy of existing techniques.This work can offer a novel technique for the automatic extensive identification of landslide boundaries in remote sensing images in addition to establishing the groundwork for future inve stigations and applications in related domains.展开更多
拥有千亿级别参数的大语言模型(large language model,LLM)已为今天的人工智能和云服务带来了巨大的技术和商业变革.然而,大模型训练与传统的通用云计算(例如,亚马逊EC2弹性计算服务)之间存在较多根本性的网络行为差异,从而带来了很多...拥有千亿级别参数的大语言模型(large language model,LLM)已为今天的人工智能和云服务带来了巨大的技术和商业变革.然而,大模型训练与传统的通用云计算(例如,亚马逊EC2弹性计算服务)之间存在较多根本性的网络行为差异,从而带来了很多新的挑战,主要包括流量模式差异造成负载难均衡(挑战1)、多训练任务通信竞争影响GPU利用率(挑战2),以及对网络故障的高敏感性(挑战3)等.因此,为通用云计算设计的数据中心网络技术(例如,网络架构、选路方法、流量调度,以及可靠性保障方法等)已不适合今天的大模型训练,这要求专门为大模型训练设计新型的数据中心网络以及配套的技术方案.介绍了阿里云专门为大模型训练设计的数据中心网络HPN以及多任务通信调度方法Crux解决上述3个挑战.HPN通过引入了一种2层、双平面(dual-plane)的网络架构,不但能够在一个Pod内高速互联15000个GPU,还能做到适用大模型训练的精准选路(解决挑战1).此外,HPN提出了一种新型的去堆叠双ToR(top-of-rack)设计来替代传统数据中心网络的单ToR交换机连接方式,根本性地避免了单点失效可靠性风险(部分解决挑战3).针对挑战2,Crux通过对GPU利用率优化问题的建模与证明,将该NP完全问题近似成GPU强度相关的流量调度问题.随后,Crux提出了一个方法优先处理具有高GPU计算强度的任务流,从而极大降低了多任务的通信竞争,优化了GPU利用率.与相关工作对比,Crux可以将GPU利用率提高多达23个百分点.HPN和Crux均已在阿里云生产环境规模化部署超过8个月,后续会持续演进迭代.在此基础上,进一步展望了大模型训练与推理领域可能的研究方向,为后续工作提供指导性建议.展开更多
文摘With the practical experience of constru-ction bidding documentation,for example,in view of the large infrastructure project construction in colleges and universities bidding documents for the main body,the construction technology,qualification,performance requirements,bill of quantities,the terms of the contract set aspects were discussed,and put forward practical measures and methods,for similar project construction bidding document preparation to provide certain reference.
文摘Investing in large transport projects affects the (potential) economic development of metropolitan areas. Yet, very little critical research has been performed to understand how to assess these effects. The relationship between infrastructure investments and regional economic development is complex and indirect, and many theoretical and methodological difficulties remain. On the one hand, the assumption that investing in infrastructure is important to sustain economic growth is sometimes doubted. On the other hand, it is argued that investments in infrastructure enhance the accessibility of urban regions and that in the slipstream of such investments, social problems in urban regions can be tackled as well. Despite these contrasting views, there is at least a consensus that transport infrastructure development depends on economic development and vice versa. Yet, in many cases, the method of assessing economic impacts highly affects the results. Therefore, this paper focuses on a critical reflection of methods for estimating economic effects of infrastructure investments. A critical evaluation is made based on Indonesian and Japanese cases. After conducting in-depth desk research on both cases, we found that the broader effects on affected group of people tend to be overlooked due to the problems of time and space dimensions, the chain reaction of effects, and inappropriate data practices. The assessment on the appraisal processes tends to overlook the broader economic implication due to narrow focus and the concept of efficiency of economic theory.
基金supported by the National Natural Science Foundation of China(Grant Nos.42090054,41931295)the Natural Science Foundation of Hubei Province of China(2022CFA002)。
文摘The frequent occurrence of extreme weather events has rendered numerous landslides to a global natural disaster issue.It is crucial to rapidly and accurately determine the boundaries of landslides for geohazards evaluation and emergency response.Therefore,the Skip Connection DeepLab neural network(SCDnn),a deep learning model based on 770 optical remote sensing images of landslide,is proposed to improve the accuracy of landslide boundary detection.The SCDnn model is optimized for the over-segmentation issue which occurs in conventional deep learning models when there is a significant degree of similarity between topographical geomorphic features.SCDnn exhibits notable improvements in landslide feature extraction and semantic segmentation by combining an enhanced Atrous Spatial Pyramid Convolutional Block(ASPC)with a coding structure that reduces model complexity.The experimental results demonstrate that SCDnn can identify landslide boundaries in 119 images with MIoU values between 0.8and 0.9;while 52 images with MIoU values exceeding 0.9,which exceeds the identification accuracy of existing techniques.This work can offer a novel technique for the automatic extensive identification of landslide boundaries in remote sensing images in addition to establishing the groundwork for future inve stigations and applications in related domains.
文摘拥有千亿级别参数的大语言模型(large language model,LLM)已为今天的人工智能和云服务带来了巨大的技术和商业变革.然而,大模型训练与传统的通用云计算(例如,亚马逊EC2弹性计算服务)之间存在较多根本性的网络行为差异,从而带来了很多新的挑战,主要包括流量模式差异造成负载难均衡(挑战1)、多训练任务通信竞争影响GPU利用率(挑战2),以及对网络故障的高敏感性(挑战3)等.因此,为通用云计算设计的数据中心网络技术(例如,网络架构、选路方法、流量调度,以及可靠性保障方法等)已不适合今天的大模型训练,这要求专门为大模型训练设计新型的数据中心网络以及配套的技术方案.介绍了阿里云专门为大模型训练设计的数据中心网络HPN以及多任务通信调度方法Crux解决上述3个挑战.HPN通过引入了一种2层、双平面(dual-plane)的网络架构,不但能够在一个Pod内高速互联15000个GPU,还能做到适用大模型训练的精准选路(解决挑战1).此外,HPN提出了一种新型的去堆叠双ToR(top-of-rack)设计来替代传统数据中心网络的单ToR交换机连接方式,根本性地避免了单点失效可靠性风险(部分解决挑战3).针对挑战2,Crux通过对GPU利用率优化问题的建模与证明,将该NP完全问题近似成GPU强度相关的流量调度问题.随后,Crux提出了一个方法优先处理具有高GPU计算强度的任务流,从而极大降低了多任务的通信竞争,优化了GPU利用率.与相关工作对比,Crux可以将GPU利用率提高多达23个百分点.HPN和Crux均已在阿里云生产环境规模化部署超过8个月,后续会持续演进迭代.在此基础上,进一步展望了大模型训练与推理领域可能的研究方向,为后续工作提供指导性建议.
