Using locally estimated geodesic distance to optimize neighborhood graph for isometric data embedding

To deal with the highly twisted and folded manifold, this paper propose a geodesic distance-based approach to build the neighborhood graph for isometric embedding. This approach assumes that the neighborhood of a point located at the highly twisted place of the manifold may not be linear so that its neighbors should be determined by geodesic distance. This approach firstly determines the neighborhood for each point using Euclidean distance and then applies the locally estimated geodesic distances to optimize the neighborhood. It increases only linear time complexity. Furthermore the optimized neighborhood can speed up the subsequent embedding process. The proposed approach is simple, general and easy to deal with a wider range of data. The conducted experiments on both synthetic and real data sets validate the approach.