Abstract:<p>The performance of most existing adaptive graph learning methods, which adjust data similarity matrix according to the data representation, depends on the hypotheses that the data representation is a good indicator of the underlying data structure. However, this hypothesis is not always applicable when dealing with high dimensional data. In this paper, we propose a novel kernel alignment unsupervised discriminative dimensionality reduction (KaUDDR) algorithm. By integrating adaptive graph learning and feature learning into a joint learning framework, graph construction and dimensionality reduction are conducted simultaneously to guarantee the optimality of graph for feature learning in the proposed algorithm. Data kernel and similarity indicator kernel are defined by learned graph and the projected data in a low-dimensional subspace, a compact and discriminative data representation in the projected subspace is obtained by means of kernel alignment to explore the consistency between the projected data kernel and similarity indicator kernel. Experimental results on dimensionality reduction as well as clustering show that our method consistently outperforms the related unsupervised dimensionality reduction algorithm.</p>

What problem does this paper attempt to address?

### Problems Addressed by the Paper This paper proposes a new unsupervised discriminative dimensionality reduction algorithm—Kernel Alignment Unsupervised Discriminative Dimensionality Reduction (KaUDDR), aiming to address several key issues in high-dimensional data dimensionality reduction: 1. **Limitations of Existing Adaptive Graph Learning Methods**: - Existing adaptive graph learning methods assume that the data representation can well reflect the underlying data structure when adjusting the data similarity matrix. However, this assumption is not always applicable when dealing with high-dimensional data. 2. **Preserving Local Intrinsic Geometric Structure During Dimensionality Reduction**: - KaUDDR ensures the simultaneous execution of graph construction and dimensionality reduction by integrating adaptive graph learning and feature learning into a unified learning framework, thereby guaranteeing the optimality of the learned graph for feature learning. 3. **Enhancing the Discriminative Ability of Dimensionality Reduction Results**: - By introducing kernel alignment technology, this algorithm can maximize the preservation of local intrinsic geometric structure in the low-dimensional subspace and make data points within different clusters more separable. 4. **Robustness and Stability**: - Due to the strong convexity and stability characteristics of kernel alignment, KaUDDR exhibits good robustness to noisy data and can achieve stable intrinsic structure representation. In summary, this paper primarily addresses how to maintain local intrinsic geometric structure and enhance the discriminative ability of dimensionality reduction results in the process of high-dimensional data dimensionality reduction through adaptive graph learning and kernel alignment technology.

Kernel alignment unsupervised discriminative dimensionality reduction

Self-representative kernel concept factorization

Discriminative Unsupervised Dimensionality Reduction.

Locality Sensitive Discriminative Unsupervised Dimensionality Reduction

Data-Dependent Kernel Discriminant Analysis for Feature Extraction and Classification

Discriminative Unsupervised 2D Dimensionality Reduction with Graph Embedding

Graph kernel by spectral transform in kernel fisher discriminant analysis

Unsupervised Adaptive Embedding for Dimensionality Reduction

Kernel Alignment for Unsupervised Feature Selection via Matrix Factorization

Kernel optimization-based discriminant analysis for face recognition

Discriminative sparse embedding based on adaptive graph for dimension reduction

Graph optimization for unsupervised dimensionality reduction with probabilistic neighbors

Joint Adaptive Graph Learning and Discriminative Analysis for Unsupervised Feature Selection

Fast and Robust Unsupervised Dimensionality Reduction with Adaptive Bipartite Graphs

Adaptive Local Linear Discriminant Analysis

Unsupervised 2D Dimensionality Reduction with Adaptive Structure Learning

Kernel-Induced Space Selection Approach to LPKHDA Dimensional Reduction Algorithm

Automatic determining optimal parameters in multi-kernel collaborative fuzzy clustering based on dimension constraint

Adaptive semi-supervised dimensionality reduction based on pairwise constraints weighting and graph optimizing

A Dimension Reduction Algorithm Preserving Both Global and Local Clustering Structure

Learning Linear Representation of Space Partitioning Trees Based on Unsupervised Kernel Dimension Reduction