Identification and anatomy of oceanic arcs within ancient orogenic belt are significant for better understanding the tectonic framework and closure process of paleo-ocean basin.This article summarizes the geological,g...Identification and anatomy of oceanic arcs within ancient orogenic belt are significant for better understanding the tectonic framework and closure process of paleo-ocean basin.This article summarizes the geological,geochemical,and geochronological characteristics of upper crust of Proto-Tethyan Lajishan intra-oceanic arc and provides new data to constrain the subduction evolution of the South Qilian Ocean.The intra-oceanic arc volcanic rocks,including intermediate-mafic lava,breccia,tuff,and minor felsic rocks,are distributed along southern part of the Lajishan ophiolite belt.Geochemical and isotopic compositions indicate that the intermediate-mafic lava were originated from depleted mantle contaminated by sediment melts or hydrous fluids,whereas the felsic rocks were likely generated by partial melting of juvenile mafic crust in intra-oceanic arc setting.Zircons from felsic rocks yield consistent and concordant ages ranging from 506 to 523 Ma,suggesting these volcanic rocks represent the relicts of upper crust of the Cambrian intra-oceanic arc.Combined with the Cambrian forearc ophiolite and accretionary complex,we suggest that the Cambrian intra-oceanic arc in the Lajishan ophiolite belt is belonging to the intra-oceanic arc system which was generated by south-directed subduction in the South Qilian Ocean at a relatively short interval between approximately 530 and 480 Ma.展开更多
个性化服务质量(Qo S,quality of service)预测是构建高质量云服务系统的重要环节,传统基于协同过滤方法采用集中式的训练模式难以保护用户隐私,为了在获取高准确预测效果的同时有效保护用户隐私,提出分布式用户隐私保护可调节的云服务...个性化服务质量(Qo S,quality of service)预测是构建高质量云服务系统的重要环节,传统基于协同过滤方法采用集中式的训练模式难以保护用户隐私,为了在获取高准确预测效果的同时有效保护用户隐私,提出分布式用户隐私保护可调节的云服务个性化QoS预测模型(DUPPA)。该模型采用“服务器-多用户”架构,服务器协调多个用户,处理多用户上传模型梯度和下载全局模型的请求并维护全局模型参数。为进一步保护用户隐私,提出用户隐私程度调节策略,通过调节本地模型参数初始化比例、梯度上传比例以平衡隐私程度和预测精度。在本地模型初始化阶段,用户计算本地模型与全局模型的差值矩阵,并选择差值矩阵中数值较大元素所对应的全局模型参数初始化本地模型参数;在梯度上传阶段,用户可选择部分重要的梯度上传至服务器来满足不同应用场景对隐私保护的需求。为了评估DUPPA的隐私程度,提出针对分布式矩阵分解模型梯度共享方案的数据重构攻击方法。实验结果表明,当DUPPA在梯度上传比例为0.1、本地模型参数初始化比例为0.5时,预测的平均绝对误差(MAE,mean absolute error)和均方根误差(RMSE,root mean square error)较传统的集中式矩阵分解模型分别降低了1.20%和0.91%;当DUPPA的梯度上传比例为0.1时,隐私程度至少是梯度上传比例为1时的5倍;当DUPPA的本地模型参数初始化比例为0.5时,隐私程度至少是本地模型参数初始化比例为1时的3.44倍。展开更多
Argo计划(Array or Real-time Geostrophic Oceanography)为海洋和大气研究提供了宝贵的资料,在短期天气预报和长期气候预测中起到了重要作用。为保证Argo观测阵列的正常运转,需要时刻关注浮标的运行状态,以保证研究区域内维持一定数量...Argo计划(Array or Real-time Geostrophic Oceanography)为海洋和大气研究提供了宝贵的资料,在短期天气预报和长期气候预测中起到了重要作用。为保证Argo观测阵列的正常运转,需要时刻关注浮标的运行状态,以保证研究区域内维持一定数量和密度的浮标。然而Argo浮标投放费用高昂,投放过早会导致资源浪费,投放过迟会导致信息资料的缺失。本文旨在使用机器学习的方法对Argo浮标在未来某个时间点的位置和状态(仍在工作或已经损坏)进行预测,以提前制定投放计划,保证在正确的位置和时间投放新的浮标,以减少资金投入。对于浮标寿命预测任务,除硬件特征之外添加额外的已存活时间作为动态属性,使用回归决策树、梯度提升回归树、随机森林和支持向量回归机等机器学习方法,对浮标剩余寿命进行预测。对于浮标轨迹预测任务,使用基于LSTM的Encoder-Decoder模型对未来多个时间步后的浮标的经/纬度信息进行预测,有效地避免了传统的LSTM模型循环单步预测所带来的误差累积问题。实验证明本文提出的浮标剩余寿命和位置预测模型都能达到较高的预测准确率,对指导浮标投放有重要意义。展开更多
基金supported by the China Geological Survey(Grant No.DD20221649)National Natural Science Foundation of China(Grant Nos.42230308,42072266)+3 种基金Bureau of Geological Exploration and Development of Qinghai Province(Grant No.[2022]32)the Xingdian Scholar Fund of Yunnan Province(Grant No.C6213001155)China Postdoctoral Science Foundation(Grant No.2021M691702)High-level Talents Project of Qinghai Province.
文摘Identification and anatomy of oceanic arcs within ancient orogenic belt are significant for better understanding the tectonic framework and closure process of paleo-ocean basin.This article summarizes the geological,geochemical,and geochronological characteristics of upper crust of Proto-Tethyan Lajishan intra-oceanic arc and provides new data to constrain the subduction evolution of the South Qilian Ocean.The intra-oceanic arc volcanic rocks,including intermediate-mafic lava,breccia,tuff,and minor felsic rocks,are distributed along southern part of the Lajishan ophiolite belt.Geochemical and isotopic compositions indicate that the intermediate-mafic lava were originated from depleted mantle contaminated by sediment melts or hydrous fluids,whereas the felsic rocks were likely generated by partial melting of juvenile mafic crust in intra-oceanic arc setting.Zircons from felsic rocks yield consistent and concordant ages ranging from 506 to 523 Ma,suggesting these volcanic rocks represent the relicts of upper crust of the Cambrian intra-oceanic arc.Combined with the Cambrian forearc ophiolite and accretionary complex,we suggest that the Cambrian intra-oceanic arc in the Lajishan ophiolite belt is belonging to the intra-oceanic arc system which was generated by south-directed subduction in the South Qilian Ocean at a relatively short interval between approximately 530 and 480 Ma.
文摘个性化服务质量(Qo S,quality of service)预测是构建高质量云服务系统的重要环节,传统基于协同过滤方法采用集中式的训练模式难以保护用户隐私,为了在获取高准确预测效果的同时有效保护用户隐私,提出分布式用户隐私保护可调节的云服务个性化QoS预测模型(DUPPA)。该模型采用“服务器-多用户”架构,服务器协调多个用户,处理多用户上传模型梯度和下载全局模型的请求并维护全局模型参数。为进一步保护用户隐私,提出用户隐私程度调节策略,通过调节本地模型参数初始化比例、梯度上传比例以平衡隐私程度和预测精度。在本地模型初始化阶段,用户计算本地模型与全局模型的差值矩阵,并选择差值矩阵中数值较大元素所对应的全局模型参数初始化本地模型参数;在梯度上传阶段,用户可选择部分重要的梯度上传至服务器来满足不同应用场景对隐私保护的需求。为了评估DUPPA的隐私程度,提出针对分布式矩阵分解模型梯度共享方案的数据重构攻击方法。实验结果表明,当DUPPA在梯度上传比例为0.1、本地模型参数初始化比例为0.5时,预测的平均绝对误差(MAE,mean absolute error)和均方根误差(RMSE,root mean square error)较传统的集中式矩阵分解模型分别降低了1.20%和0.91%;当DUPPA的梯度上传比例为0.1时,隐私程度至少是梯度上传比例为1时的5倍;当DUPPA的本地模型参数初始化比例为0.5时,隐私程度至少是本地模型参数初始化比例为1时的3.44倍。
文摘Argo计划(Array or Real-time Geostrophic Oceanography)为海洋和大气研究提供了宝贵的资料,在短期天气预报和长期气候预测中起到了重要作用。为保证Argo观测阵列的正常运转,需要时刻关注浮标的运行状态,以保证研究区域内维持一定数量和密度的浮标。然而Argo浮标投放费用高昂,投放过早会导致资源浪费,投放过迟会导致信息资料的缺失。本文旨在使用机器学习的方法对Argo浮标在未来某个时间点的位置和状态(仍在工作或已经损坏)进行预测,以提前制定投放计划,保证在正确的位置和时间投放新的浮标,以减少资金投入。对于浮标寿命预测任务,除硬件特征之外添加额外的已存活时间作为动态属性,使用回归决策树、梯度提升回归树、随机森林和支持向量回归机等机器学习方法,对浮标剩余寿命进行预测。对于浮标轨迹预测任务,使用基于LSTM的Encoder-Decoder模型对未来多个时间步后的浮标的经/纬度信息进行预测,有效地避免了传统的LSTM模型循环单步预测所带来的误差累积问题。实验证明本文提出的浮标剩余寿命和位置预测模型都能达到较高的预测准确率,对指导浮标投放有重要意义。