基于可分类网络的图像识别新方法

提出一种新的基于可分类网络的图像识别方法,从一种新的可分类网络的视角重新审视平面灰度图像,将图像灰度矩阵视为可分类网络的连接关系矩阵,通过Hadamard乘积变换和负连接关系可分类网络的拓扑结构,产生新的图像识别特征参量对图像进行描述。该方法具有高速和高识别率的特性。

基于洞主的广义符号网络拓扑结构分析

针对广义符号网络,论文在复杂网络结构洞、洞主、无特权网络等概念的基础上,提出了节点的邻居集和洞数的概念,并基于这两个概念对无特权网络进行分析。之后,考虑了与无特权网络无洞主相反的情况,即每个节点都是洞主的网络,提出了全洞主网络的概念,并给出了一种生成全洞主网络的方法,对这样的网络进行了分析并将其与无特权网络进行了比较。最后,数值仿真验证了提出的方法的有效性。

Classification of frozen soil blastability by using perceptron neural network

Influence factors of frozen soil blastability are analyzed which mainly conclude the strain energy coefficient, tensile strength, compressive strength, longitudinal wave velocity and transverse wave velocity. According to the principle of perceptron neural network, at first the index factors are standardized by the aid of the efficient function theory, then the blastability of frozen sand at -7, -12 and -17 ℃ are classified three categories. Through adjusting the weight value and threshold value, we can obtain that the clay blastability at -7 ℃ is close to the sand blastability at -12 ℃, they belong to the second category, the clay blastability at -12 ℃ is close to the sand blastability at -17 ℃, thus they are divided into the third category.