摘要
介绍了流形学习中Hessian特征映射、拉普拉斯特征映射和局部切空间排列3种非线性降维算法的概念和实现步骤,并基于三维的Swiss Roll数据点集通过实验对3种算法在参数选择和运算效率等方面进行了比较分析,期望为不同应用提供参考.
The concept and steps to achieve for three nonlinear dimensionality reduction algorithms of hessian eigenmaps, laplacian eigenmaps and local tangent space alignment were introduced in manifold learning, as well as comparison and analysis of these three algorithms on parameters selection and operation efficiency were given based on three-dimensional Swiss Rolldata point collection through experiments, it is expected to give some insight into different applications.
出处
《云南民族大学学报(自然科学版)》
CAS
2009年第2期151-156,共6页
Journal of Yunnan Minzu University:Natural Sciences Edition
基金
国家自然科学基金重大资助项目(50490270)
国家杰出青年科学基金资助项目(50225414)