Nonlinear dimensionality reduction of manifolds by diffusion maps
Song Yi-mei
Abstract
Song Yi-mei
Abstract
Nonlinear dimensionality reduction programs keep the local properties but relax the distances between points which are not in a neighborhood.As a new learning framework,the diffusion method realizes dimensionality reduction in a diffusion processing.Based on the theory of diffusion maps,this paper discusses the numerical method for spectral decomposition and presents the diffusion maps algorithm(DMA).Experimental results show that the DMA technique can detect the intrinsic dimensionality in high-dimensional data and is more stable in noise case.
OpenAlex reports 1 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
Nonlinear dimensionality reduction programs keep the local properties but relax the distances between points which are not in a neighborhood.As a new learning framework,the diffusion method realizes dimensionality reduction in a diffusion processing.Based on the theory of diffusion maps,this paper discusses the numerical method for spectral decomposition and presents the diffusion maps algorithm(DMA).Experimental results show that the DMA technique can detect the intrinsic dimensionality in high-dimensional data and is more stable in noise case.
Key concepts: Diffusion map, Dimensionality reduction, Curse of dimensionality, Diffusion, Nonlinear dimensionality reduction, Nonlinear system, Reduction (mathematics), Computer science