Similarity Consideration for Visualization and Manifold Geometry Preservation

Shashwati Mishra, Chittaranjan Pradhan
2011 International Journal of Computer Applications  
Manifold learning techniques are used to preserve the original geometry of dataset after reduction by preserving the distance among data points. MDS (Multidimensional Scaling), ISOMAP (Isometric Feature Mapping), LLE (Locally Linear Embedding) are some of the geometrical structure preserving dimension reduction methods. In this paper, we have compared MDS and ISOMAP and considered similarity as an approach to find the reduced representation of original data using ISOMAP.
doi:10.5120/1981-2664 fatcat:mawi3aw4unajplhsf4ov25isjy