2010•Journal of Xidian UniversityRequires access

Nonlinear dimensionality reduction of manifolds by diffusion maps

Song Yi-mei

Open publisher page 1 citations

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.

About this research paper

What this paper is about

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.

Why it matters

OpenAlex reports 1 citations for this work. Citation counts describe recorded attention and do not establish research quality.

Key contribution

A contribution statement is not available in the OpenAlex record.

Method / approach

Method details are not available in the OpenAlex metadata.

Main findings

Findings are not separately available in the OpenAlex metadata.

Limitations

Limitations are not available in the OpenAlex metadata.

Applications

Application details are not available in the OpenAlex metadata.

Available 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.

Key concepts: Diffusion map, Dimensionality reduction, Curse of dimensionality, Diffusion, Nonlinear dimensionality reduction, Nonlinear system, Reduction (mathematics), Computer science

Related papers

Back to paper searchBrowse research topicsOriginal source
Nonlinear dimensionality reduction of manifolds by diffusion maps — Research Paper | ScholarLens