Abstract:Dimensional reduction (DR) maps high-dimensional data into a lower dimensions latent space with minimized defined optimization objectives. The two independent branches of DR are feature selection (FS) and feature projection (FP). FS focuses on selecting a critical subset of dimensions but risks destroying the data distribution (structure). On the other hand, FP combines all the input features into lower dimensions space, aiming to maintain the data structure, but lacks interpretability and sparsity. Moreover, FS and FP are traditionally incompatible categories and have not been unified into an amicable framework. Therefore, we consider that the ideal DR approach combines both FS and FP into a unified end-to-end manifold learning framework, simultaneously performing fundamental feature discovery while maintaining the intrinsic relationships between data samples in the latent space. This paper proposes a unified framework named Unified Dimensional Reduction Network (UDRN) to integrate FS and FP in an end-to-end way. Furthermore, a novel network framework is designed to implement FS and FP tasks separately using a stacked feature selection network and feature projection network. In addition, a stronger manifold assumption and a novel loss function are proposed. Furthermore, the loss function can leverage the priors of data augmentation to enhance the generalization ability of the proposed UDRN. Finally, comprehensive experimental results on four image and four biological datasets, including very high-dimensional data, demonstrate the advantages of DRN over existing methods (FS, FP, and FS&FP pipeline), especially in downstream tasks such as classification and visualization.

Nonnegative Representation Based Discriminant Projection for Face Recognition.

Two-stage nonnegative sparse representation for large-scale face recognition.

UDRN: Unified Dimensional Reduction Neural Network for Feature Selection and Feature Projection

Learning a Locality Preserving Subspace for Visual Recognition.

UDRN: Unified Dimensional Reduction Neural Network for Feature Selection and Feature Projection.

Combination of Local Multiple Patterns and Exponential Discriminant Analysis for facial recognition

Local similarity preserving projections for face recognition

Local Feature Discriminant Projection

Two Phase Representation Discriminant Projection for Face Recognition

Non-negative representation based discriminative dictionary learning for face recognition

Face Feature Extraction and Recognition Via Local Binary Pattern and Two-Dimensional Locality Preserving Projection

Two Dimensional Discriminant Neighborhood Preserving Embedding in Face Recognition

Collaborative Discriminant Locality Preserving Projections With its Application to Face Recognition

Enhanced Face Recognition Using Tensor Neighborhood Preserving Discriminant Projections

Minimum-distance discriminant projection and its application to face recognition

By Normalizing To Improve Generalized Foley-Sammon Transform In High-Dimensional Spaces-----With Application To Face Recognition

Discriminant Maximum Margin Projections for Face Recognition.

Nonparametric Locality Reforming Projection for Linear Discriminative Dimensionality Reduction

Multi-layer Cross-domain Non-negative Matrix Factorization for Cross-scene Dimension Reduction on Hyperspectral Images.

Collaborative Representation Based Local Discriminant Projection for Feature Extraction

Low-rank Representation Based Discriminative Projection for Robust Feature Extraction