返回
Towards robust and sparse linear discriminant analysis for image classification
DOI:10.1016/j.patcog.2024.110512.png)
摘要
En 中文
Linear discriminant analysis (LDA) is a popular dimensionality reduction technique that has been widely used in pattern recognition. However, there exist a large number of redundant features and corrupted noise in realworld applications, which makes the performance of existing LDA methods degrade and thus leads to a decrease in classification accuracy. To address the above issues, we propose a novel robust and sparse LDA formulation dubbed RSLDA+. The key idea is introducing the mixed sparse regularization, i.e., e 0 -norm plus e 2 , 0 -norm, for feature representation and enforce e 0 -norm for noise reduction. Furthermore, an optimization algorithm based on the alternating direction method of multipliers (ADMM) is developed in combination with hard thresholding operators. Extensive experiments on six common image datasets verify that the proposed RSLDA+ outperforms state-of-the-art LDA variants in classification accuracy. In addition, the ablation, robustness, convergence, stability, and sparsity are analyzed in detail. The results suggest that the proposed RSLDA+ provides an effective and robust method for image classification.
Keyword:
Image classification
Linear discriminant analysis (LDA)
Sparse representation
e0-norm
期刊
IF:
7.6
论文数:
1.3W
被引数:
4.5W
机构
引用论文
Linear discriminant analysis with generalized kernel constraint for robust image classification
PATTERN RECOGNITION
IF7.6
A Selective Overview of Sparse Principal Component Analysis稀疏主成分分析的选择性综述
PROCEEDINGS OF THE IEEE
IF25.9
Robust and Sparse Linear Discriminant Analysis via an Alternating Direction Method of Multipliers通过乘数的交替方向方法进行鲁棒和稀疏的线性判别分析
Expression in Brain of a Messenger RNA Encoding a Novel Neuropeptide Homologous to Calcitonin Gene-Related Peptide
Science
IF0

