Under the requirement of everything over IP, network service shows the following characteristics:(1) network service increases its richness;(2) broadband streaming media becomes the mainstream. To achieve unified mult...Under the requirement of everything over IP, network service shows the following characteristics:(1) network service increases its richness;(2) broadband streaming media becomes the mainstream. To achieve unified multi-service bearing in the IP network, the largescale access convergence network architecture is proposed. This flat access convergence structure with ultra-small hops, which shortens the service transmission path, reduces the complexity of the edge of the network, and achieves IP strong waist model with the integration of computation, storage and transmission. The key technologies are also introduced in this paper, including endto-end performance guarantee for real time interactive services, fog storing mechanism, and built-in safety transmission with integration of aggregation and control.展开更多
Three recent breakthroughs due to AI in arts and science serve as motivation:An award winning digital image,protein folding,fast matrix multiplication.Many recent developments in artificial neural networks,particularl...Three recent breakthroughs due to AI in arts and science serve as motivation:An award winning digital image,protein folding,fast matrix multiplication.Many recent developments in artificial neural networks,particularly deep learning(DL),applied and relevant to computational mechanics(solid,fluids,finite-element technology)are reviewed in detail.Both hybrid and pure machine learning(ML)methods are discussed.Hybrid methods combine traditional PDE discretizations with ML methods either(1)to help model complex nonlinear constitutive relations,(2)to nonlinearly reduce the model order for efficient simulation(turbulence),or(3)to accelerate the simulation by predicting certain components in the traditional integration methods.Here,methods(1)and(2)relied on Long-Short-Term Memory(LSTM)architecture,with method(3)relying on convolutional neural networks.Pure ML methods to solve(nonlinear)PDEs are represented by Physics-Informed Neural network(PINN)methods,which could be combined with attention mechanism to address discontinuous solutions.Both LSTM and attention architectures,together with modern and generalized classic optimizers to include stochasticity for DL networks,are extensively reviewed.Kernel machines,including Gaussian processes,are provided to sufficient depth for more advanced works such as shallow networks with infinite width.Not only addressing experts,readers are assumed familiar with computational mechanics,but not with DL,whose concepts and applications are built up from the basics,aiming at bringing first-time learners quickly to the forefront of research.History and limitations of AI are recounted and discussed,with particular attention at pointing out misstatements or misconceptions of the classics,even in well-known references.Positioning and pointing control of a large-deformable beam is given as an example.展开更多
Grouping nodes gives better performance to the whole network by diminishing the average network delay and avoiding unnecessary message forwarding and additional overhead. Many routing protocols for ad-hoc and sensor n...Grouping nodes gives better performance to the whole network by diminishing the average network delay and avoiding unnecessary message forwarding and additional overhead. Many routing protocols for ad-hoc and sensor networks have been designed but none of them are based on groups. In this paper, we will start defining group-based topologies, and then we will show how some wireless ad hoc sensor networks (WAHSN) routing protocols perform when the nodes are arranged in groups. In our proposal connections between groups are established as a function of the proximity of the nodes and the neighbor's available capacity (based on the node's energy). We describe the architecture proposal, the messages that are needed for the proper operation and its mathematical description. We have also simulated how much time is needed to propagate information between groups. Finally, we will show a comparison with other architectures.展开更多
首先,基于云计算应用模式,提出一种能有效利用云存储架构的双层缓存技术.通过在客户端和服务器端建立分布式缓存,能有效避免用户频繁访问远端数据,为用户构建轻量级的客户端,解决了目前地学数据可视化软件大量占用用户本地存储容量的问...首先,基于云计算应用模式,提出一种能有效利用云存储架构的双层缓存技术.通过在客户端和服务器端建立分布式缓存,能有效避免用户频繁访问远端数据,为用户构建轻量级的客户端,解决了目前地学数据可视化软件大量占用用户本地存储容量的问题.同时服务器端也避免了多次访问云存储文件系统,减少了大量的数据检索与加载时间.其次,提出一种ARLS(association rule last successor)访问预测算法,根据用户的历史访问记录,利用关联规则挖掘用户的访问模式,对其访问行为进行预测,进而提前加载数据,提高缓存命中率,解决了用户在可视化过程中不断移动兴趣区域,频繁更换渲染数据的问题,能有效应对用户具有多种访问模式的情况,提高了预测准确率.实验结果表明,该云存储架构显著减少了本地资源消耗,访问预测算法的准确率在最差情形下可达47.59%,平均准确率达91.3%,分布式缓存的平均缓存命中率达95.61%,可有效支持云端大规模地震数据的快速可视化.展开更多
基金supported by The National Key Technology R&D Program (Grant No. 2011BAH19B00)The National Basic Research Program of China (973) (Grant No. 2012CB315900)The National High Technology Research and Development Program of China (863) (Grant No. 2015AA016102)
文摘Under the requirement of everything over IP, network service shows the following characteristics:(1) network service increases its richness;(2) broadband streaming media becomes the mainstream. To achieve unified multi-service bearing in the IP network, the largescale access convergence network architecture is proposed. This flat access convergence structure with ultra-small hops, which shortens the service transmission path, reduces the complexity of the edge of the network, and achieves IP strong waist model with the integration of computation, storage and transmission. The key technologies are also introduced in this paper, including endto-end performance guarantee for real time interactive services, fog storing mechanism, and built-in safety transmission with integration of aggregation and control.
文摘Three recent breakthroughs due to AI in arts and science serve as motivation:An award winning digital image,protein folding,fast matrix multiplication.Many recent developments in artificial neural networks,particularly deep learning(DL),applied and relevant to computational mechanics(solid,fluids,finite-element technology)are reviewed in detail.Both hybrid and pure machine learning(ML)methods are discussed.Hybrid methods combine traditional PDE discretizations with ML methods either(1)to help model complex nonlinear constitutive relations,(2)to nonlinearly reduce the model order for efficient simulation(turbulence),or(3)to accelerate the simulation by predicting certain components in the traditional integration methods.Here,methods(1)and(2)relied on Long-Short-Term Memory(LSTM)architecture,with method(3)relying on convolutional neural networks.Pure ML methods to solve(nonlinear)PDEs are represented by Physics-Informed Neural network(PINN)methods,which could be combined with attention mechanism to address discontinuous solutions.Both LSTM and attention architectures,together with modern and generalized classic optimizers to include stochasticity for DL networks,are extensively reviewed.Kernel machines,including Gaussian processes,are provided to sufficient depth for more advanced works such as shallow networks with infinite width.Not only addressing experts,readers are assumed familiar with computational mechanics,but not with DL,whose concepts and applications are built up from the basics,aiming at bringing first-time learners quickly to the forefront of research.History and limitations of AI are recounted and discussed,with particular attention at pointing out misstatements or misconceptions of the classics,even in well-known references.Positioning and pointing control of a large-deformable beam is given as an example.
文摘Grouping nodes gives better performance to the whole network by diminishing the average network delay and avoiding unnecessary message forwarding and additional overhead. Many routing protocols for ad-hoc and sensor networks have been designed but none of them are based on groups. In this paper, we will start defining group-based topologies, and then we will show how some wireless ad hoc sensor networks (WAHSN) routing protocols perform when the nodes are arranged in groups. In our proposal connections between groups are established as a function of the proximity of the nodes and the neighbor's available capacity (based on the node's energy). We describe the architecture proposal, the messages that are needed for the proper operation and its mathematical description. We have also simulated how much time is needed to propagate information between groups. Finally, we will show a comparison with other architectures.
文摘首先,基于云计算应用模式,提出一种能有效利用云存储架构的双层缓存技术.通过在客户端和服务器端建立分布式缓存,能有效避免用户频繁访问远端数据,为用户构建轻量级的客户端,解决了目前地学数据可视化软件大量占用用户本地存储容量的问题.同时服务器端也避免了多次访问云存储文件系统,减少了大量的数据检索与加载时间.其次,提出一种ARLS(association rule last successor)访问预测算法,根据用户的历史访问记录,利用关联规则挖掘用户的访问模式,对其访问行为进行预测,进而提前加载数据,提高缓存命中率,解决了用户在可视化过程中不断移动兴趣区域,频繁更换渲染数据的问题,能有效应对用户具有多种访问模式的情况,提高了预测准确率.实验结果表明,该云存储架构显著减少了本地资源消耗,访问预测算法的准确率在最差情形下可达47.59%,平均准确率达91.3%,分布式缓存的平均缓存命中率达95.61%,可有效支持云端大规模地震数据的快速可视化.