期刊文献+
共找到4篇文章
< 1 >
每页显示 20 50 100
基于分区层次图的海量高维数据学习索引构建方法
1
作者 华悦琳 周晓磊 +2 位作者 范强 王芳潇 严浩 《计算机工程与科学》 CSCD 北大核心 2024年第7期1193-1201,共9页
学习索引是破解海量高维数据近似最近邻搜索问题的关键。然而,现有学习索引技术结果仅局限于单个分区中,且依赖于近邻图的构建。随着数据维度和规模的增长,索引难以对分区边界数据进行精确判断,并且构建时间复杂度增大,可扩展性难以保... 学习索引是破解海量高维数据近似最近邻搜索问题的关键。然而,现有学习索引技术结果仅局限于单个分区中,且依赖于近邻图的构建。随着数据维度和规模的增长,索引难以对分区边界数据进行精确判断,并且构建时间复杂度增大,可扩展性难以保障。针对上述问题,提出了基于分区层次图的学习索引方法PBO-HNSW。该方法对分区边界数据进行重新分配,并行构建分布式图索引结构,从而有效应对近似最近邻搜索问题所面临的挑战。实验结果表明,该方法能够在百万级海量高维数据上实现毫秒级的索引构建。当召回率为0.93时,PBO-HNSW方法构建时间仅为基线方法的36.4%。 展开更多
关键词 近似最近邻搜索 学习索引 层次可导航小世界图 分区学习 索引结构
下载PDF
Fiscal Decentralization and Economic Growth in China: A Meta-Regression Analysis
2
作者 谢贞发 张玮 《China Economist》 2016年第5期28-42,共15页
In order to test whether the major empirical results on the "relationship between fiscal decentralization and economic growth in China" are affected by study characteristics, this paper conducts a meta-analysis of t... In order to test whether the major empirical results on the "relationship between fiscal decentralization and economic growth in China" are affected by study characteristics, this paper conducts a meta-analysis of the major existing empirical literature. Our analysis indicates that some empirical results on how China's jqscal decentralization affects economic growth are subject to different study characteristics. In particular, empirical results that fiscal decentralization has "significant positive effect" on economic growth are subject to such study characteristics as "region, labor and capital growth rate, other reforms and intra-budget capital." Through the funnel plot asymmetry test, the problem of publication bias is found to exist in the sampled literature and is concentrated in spending decentralization. 展开更多
关键词 fiscal decentralization economic growth META-ANALYSIS
下载PDF
Study of Preservation Status and Dietary Reconstruction in the Human Remains Recovered from Roopkund Lake through Chemical Analysis of Faunal Remains
3
作者 Yogambar Singh Farswan Jaibir Singh 《Journal of Chemistry and Chemical Engineering》 2015年第1期15-22,共8页
Present study is carried out in the bone samples collected from Roopkund Lake in district Chamoli Garhwal, Uttarakhand which is located at 5,029 meters from main sea level in between Nanda Ghunghti and Trishuli peak. ... Present study is carried out in the bone samples collected from Roopkund Lake in district Chamoli Garhwal, Uttarakhand which is located at 5,029 meters from main sea level in between Nanda Ghunghti and Trishuli peak. This historical site belongs to 9th century A.D. All the samples selected for the study were dried in room temperature as well as hot air oven at 32 ~C. Cleaning, pretreatment and digestion process of faunal remains was followed through established scientific methods. Chemical analysis i.e. concentration of different elements such as calcium, strontium, barium, magnesium and zinc as well as isotopic ratios of Carbon and Nitrogen was estimated with the help of ICP (inductively coupled plasma spectroscopy) and AAS (atomic absorption spectrophotometer). The results obtained from the chemical analysis are significant. On the basis of concentration of different elements and ratios of Nitrogen and Carbon isotopes, the dietary habits of the peoples buried in the Roopkund Lake are identified, which is different from sample to sample person to person. Besides this, the results are also significantly helpful for knowing the preservation status of faunal remains in Roopkund Lake. Finally this study also indicated the potentiality of chemical analysis for reconstructing the palaeodiet behaviour and preservation status of bone remains. 展开更多
关键词 Roopkund Lake dietary reconstruction chemical analysis faunal remains archaeological site.
下载PDF
A NEW NEURAL NETWORK-BASED ADAPTIVE ILC FOR NONLINEAR DISCRETE-TIME SYSTEMS WITH DEAD ZONE SCHEME 被引量:2
4
作者 Ronghu CHI Zhongsheng HOU 《Journal of Systems Science & Complexity》 SCIE EI CSCD 2009年第3期435-445,共11页
By introducing a deadwzone scheme, a new neural network based adaptive iterative learning control (ILC) (NN-AILC) scheme is presented for nonlinear discrete-time systems, where the NN weights are time-varying. The... By introducing a deadwzone scheme, a new neural network based adaptive iterative learning control (ILC) (NN-AILC) scheme is presented for nonlinear discrete-time systems, where the NN weights are time-varying. The most distinct contribution of the proposed NN-AILC is the relaxation of the identical conditions of initial state and reference trajectory, which are common requirements in traditional ILC problems. Convergence analysis indicates that the tracking error converges to a bounded ball, whose size is determined by the dead-zone nonlinearity. Computer simulations verify the theoretical results. 展开更多
关键词 Adaptive control iterative learning control neural network non-identical initial condition non-identical trajectory.
原文传递
上一页 1 下一页 到第
使用帮助 返回顶部