BAM网络及其在结构分析中的应用

基于BAM网络、有限元结构分析能量函数的相似性，建立了一种新的结构分析方法。 表明，BAM网络能有效地完成有限元模型求解。     

Robustness of Cooperation on Highly Clustered Scale-Free Networks

We study the effect of mutation on the evolutionary prisoner＇s dilemma in highly clustered scale-free networks. It is found that cooperation is more sensitive and vulnerable to strategy mutation in more highly clustered networks. For small mutation rates, high clustering coefficient promotes cooperation. For medium mutation rates, high clustering coefficient inhibits the emergence of cooperation. For large mutation rates, cooperation is insensitive to clustering property. We provide explanations for the effects of clustering on cooperation with varied mutation rates.     

Key module for a novel all-optical network coding scheme

<正>We introduce a novel network-coding scheme that can be implemented in all-optical multicast networks.A simple and successful module based on the all-optical XOR gate is designed to realize the network coding scheme.The module is a key hardware component in realizing the proposed scheme.The working principle and the experimental results of the module are also presented.Experimental results show that the function of the module is sufficient in satisfying the requirements of the proposed network coding scheme.     

STABILITY OF THE LCD MODEL

The growing network model with loops and multiple edges proposed by Bolloba′s et al.(Random Structures and Algorithms 18(2001)) is restudied from another perspective.Based on the first-passage probability of Markov chains,we prove that the degree distribution of the LCD model is power-law with degree exponent 3 as the network size grows to infinity.     

基于云模型和PageRank算法的社会网络群决策方法

近年来，随着信息、通信和技术的快速发展，决策者之间的关系也由此变得越来越紧密。在此背景下，决策者基于社会网络关系存在交互，社会网络群体决策会产生比传统群体决策更多的信息，信息不完全性、随机性以及决策者有限理性等因素都会导致群体决策结果与实际情况产生偏差。针对社会网络群体决策中决策偏好信息为定性概念的语言信息，且决策者之间存在信任度的群体决策问题，本文提出了一种基于云模型和PageRank算法的社会网络群决策方法并针对现有的云聚类算法拓展出了社会网络云聚类算法。首先，将语言信息转化为云模型；其次，基于社会网络拓扑图利用PageRank算法来计算每个决策者的权重；接着，使用社会网络云聚类算法将决策者分成几个子聚集并求出子聚集的权重；最后把云模型综合成方案云，利用随机模拟技术获取各个方案云的评分，给出方案的排序，并通过对比分析验证了该方法的有效性。     

LK-CAUNet：基于交叉注意的大内核多尺度可变形医学图像配准网络

经典的UNet网络可用于预测全分辨率空间域的密集位移场，在医学图像配准中取得了巨大成功。但对大变形的三维图像配准，还存在运行时间长、无法有效保持拓扑结构、空间特征易丢失等缺点。为此，提出一种基于交叉注意的大内核多尺度可变形医学图像配准网络（large kernel multi-scale deformable medical image registration network based on cross-attention,LK-CAUNet）。在经典UNet模型基础上，通过引入交叉注意力模块，实现高效、多层次的语义特征融合；配备大内核非对称并行卷积，使其具有多尺度特征和对复杂结构的学习能力；通过加入平方和缩放模块，实现拓扑守恒和变换可逆。基于脑部MRI数据集，将LK-CAUNet与18种经典图像配准模型进行了比较，结果表明，LK-CAUNet的配准性能较其他模型有明显提升，其Dice得分较TransMorph配准方法提高了8%，而参数量仅为TransMorph的1/5。