Pre-process of computational model used for numerical manifold method is usually difficult in practical applications.
数值流形方法的前处理一直是一个难题。In this paper, a few manifold learning algorithms are improved and used in the target recognition of millimeter-wave (MMW) detector.
本文将流形学习方法应用于毫米波探测器目标识别,并对现有流形学习算法进行了改进和推广。As a new learning framework, the diffusion method realizes dimensionality reduction in a diffusion processing.
作为一种新的流形学习框架,扩散映射通过在扩散过程中保持扩散距离进行降维。The hidden manifold in the high dimensional space can be successfully embedded to a low dimensional space using KPCA.
核主元分析是一种非线性降维算法,能够把这种流形结构嵌入到低维空间。The Locally Linear Embedding(LLE) algorithm is an effective technique for nonlinear dimensionality reduction of high-dimensional data.
局部线性嵌入(LLE)算法是有效的非线性降维方法,时间复杂度低并具有强的流形表达能力。Researched results show that the manifold method is effective for the simulation of the dynamic failure process under shock loading.
从而验证了流形元在模拟冲击载荷作用下材料动态破坏过程的有效性和可行性。The arithmetic will also be generalized to other designed manifold, so the stability and anti-disturbance ability will be improved.
该算法可以进一步推广到其他各种设计流形,从而提高磁悬浮系统的稳定性与抗干扰性。Within differential geometry are the concepts of fiber bundles and calculus on manifolds, in particular, vector and tensor calculus.
微分几何包含这样的概念:纤维束和流形上的微积分,特别是矢量与张量微积分。The likelihood is fused with prior human poses in a Bayesian framework to specify the coordinates of the pose in the manifold space.
通过融合似然概率和人体姿态的先验概率,利用Bayes推理框架就可以求出输入的人体姿态在流形空间的坐标。