Abstract:Images are often represented as vectors with high dimensions when involved in classification. As a result, dimensionality reduction methods have to be developed to avoid the curse of dimensionality. Among them, Laplacian eigenmaps (LE) have attracted widespread concentrations. In the original LE, point to point (P2P) distance metric is often adopted for manifold learning. Unfortunately, they show few impacts on robustness to noises. In this paper, a novel supervised dimensionality reduction method, named feature space to feature space distance metric learning (FSDML), is presented. For any point, it can construct a feature space spanned by its k intra-class nearest neighbors, which results in a local projection on its nearest feature space. Thus feature space to feature space (S2S) distance metric will be defined to Euclidean distance between two corresponding projections. On one hand, the proposed S2S distance metric displays superiority on robustness by the local projection. On the other hand, the projection on the nearest feature space contributes to fully mining local geometry information hidden in the original data. Moreover, both class label similarity and dissimilarity are also measured, based on which an intra-class graph and an inter-class graph will be individually modeled. Finally, a subspace can be found for classification by maximizing S2S based manifold to manifold distance and preserving S2S based locality of manifolds, simultaneously. Compared to some state-of-art dimensionality reduction methods, experiments validate the proposed method's performance either on synthesized data sets or on benchmark data sets.

Dimensionality Reduction Algorithm Based on Density Portrayal

Towards mining the density and discriminative information of data in dimensionality reduction

Dimension Reduction Algorithm Based on Locality and Discriminant Characteristics

L-PCA-based dimensionality reduction algorithm for high dimension images

Probability-Based Locally Linear Embedding for Classification

Dimensionality Reduction by Locally Linear Discriminant Analysis for Handwritten Chinese Character Recognition.

Supervised locally linear embedding with probability-based distance for classification

A High Accuracy Nonlinear Dimensionality Reduction Optimization Method.

Adaptive Local Linear Discriminant Analysis

Optimal Dimensionality Discriminant Analysis and Its Application to Image Recognition

Adaptive and Fuzzy Locality Discriminant Analysis for Dimensionality Reduction

Discriminant-enhanced neighborhood preserving embedding for dimensionality reduction

Linear discriminant regression and local discriminant analysis based dimensionality reduction algorithm

Robust dimensionality reduction via feature space to feature space distance metric learning.

Dimensionality reduction by low-rank embedding

Divergent Projection Analysis for Unsupervised Dimensionality Reduction

Constrained discriminant neighborhood embedding for high dimensional data feature extraction.

An embedding dimension reduction algorithm based on sparse analysis

Dimensionality Reduction Via Preserving Local Information

Local Regularized Least-Square Dimensionality Reduction

Relative Distance-Based Laplacian Eigenmaps