As a new promising paradigm, cloud computing can make good use of economics of scale and elastically deliver almost any IT related services on demand. Nevertheless, one of the key problems remaining in cloud computing...As a new promising paradigm, cloud computing can make good use of economics of scale and elastically deliver almost any IT related services on demand. Nevertheless, one of the key problems remaining in cloud computing is related to virtual machine images, which require a great amount of space/time to reposit/provision, especially with diverse requests from thousands of users simultaneously. In this paper, by using the splitting and eliminating redundant data techniques, a space and time efficient approach for virtual machines is proposed. The experiments demonstrate that, compared with existing solutions, our approach can conserve more disk space and speed up the provisioning of virtual machines.展开更多
科学评估地下空间开发需求潜力是缓解城市化问题和合理拓展有限区域的重要基础工作。目前地下空间评价中的社会经济数据多来自于传统官方文件,其全面完整性和时空精度并不理想;此外主客观赋权方法的使用,一定程度上存在主观性强和受数...科学评估地下空间开发需求潜力是缓解城市化问题和合理拓展有限区域的重要基础工作。目前地下空间评价中的社会经济数据多来自于传统官方文件,其全面完整性和时空精度并不理想;此外主客观赋权方法的使用,一定程度上存在主观性强和受数据干扰等不足。文章以多源大数据支持的指标体系为基础,构建熵权-随机森林耦合的地下空间需求评价模型。该模型基于熵权法确定负样本,将总样本和指标因子导入随机森林算法中,挖掘社会经济指标与现有地下设施间的复杂非线性关系。研究表明,经过网格搜索调优后的模型AUC(area under curve)精度达到0.979,其中77.45%的现有设施落入评价的高需求区内,证明所采用模型有较强的准确性和可靠性,其精细化评价结果可为今后地下建设选址提供更符合实际的借鉴。展开更多
基金Project supported by the Shanghai Leading Academic Discipline Project(Grant No.J50103)the Natural Science Foundation of Shanghai Municipality(Grant No.10Z1411600)+1 种基金the Innovation Foundation of Shanghai Municipal Education Commission(Grant No.10YZ18)the National Science and Technology Major Project(Grant No.LX101102103)
文摘As a new promising paradigm, cloud computing can make good use of economics of scale and elastically deliver almost any IT related services on demand. Nevertheless, one of the key problems remaining in cloud computing is related to virtual machine images, which require a great amount of space/time to reposit/provision, especially with diverse requests from thousands of users simultaneously. In this paper, by using the splitting and eliminating redundant data techniques, a space and time efficient approach for virtual machines is proposed. The experiments demonstrate that, compared with existing solutions, our approach can conserve more disk space and speed up the provisioning of virtual machines.
文摘科学评估地下空间开发需求潜力是缓解城市化问题和合理拓展有限区域的重要基础工作。目前地下空间评价中的社会经济数据多来自于传统官方文件,其全面完整性和时空精度并不理想;此外主客观赋权方法的使用,一定程度上存在主观性强和受数据干扰等不足。文章以多源大数据支持的指标体系为基础,构建熵权-随机森林耦合的地下空间需求评价模型。该模型基于熵权法确定负样本,将总样本和指标因子导入随机森林算法中,挖掘社会经济指标与现有地下设施间的复杂非线性关系。研究表明,经过网格搜索调优后的模型AUC(area under curve)精度达到0.979,其中77.45%的现有设施落入评价的高需求区内,证明所采用模型有较强的准确性和可靠性,其精细化评价结果可为今后地下建设选址提供更符合实际的借鉴。
