A Manifold-Based Dimension Reduction Algorithm Framework for Noisy Data Using Graph Sampling and Spectral Graph

Figure 1

(a) Diagram for calculating the deviation value of each dimension in the dimension reduction result of a manifold with distribution parameters using the anchor vector and its locality as an example. (b) Assuming that is located in the local range of two anchor vectors and , this diagram calculates the deviation values of the dimension reduction result of .

(a)

(b)