A Sparse Fixed-Point Online KPCA Extraction Algorithm
- 20 August 2024
- journal article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Signal Processing
- Vol. 72 (1053587X) , 4604-4617
Abstract
Kernel principal component analysis (KPCA) is a powerful nonlinear feature extraction tool, but its canonical formulation is inadequate for streaming data. Despite the existence of some efficient online KPCA solutions, genuinely sparse online KPCA algorithms are missing in the literature. This paper presents a new, fast, and accurate online fixed-point algorithm for sparse kernel principal component extraction. By exploring a two-level sparsifying strategy, the proposed algorithm can efficiently manage streaming data and large datasets within small computing and memory footprints, achieving higher accuracy and sparser components than alternative online KPCA techniques.Keywords
This publication has 30 references indexed in Scilit:
- Sparse Kernel Principal Component Analysis via Sequential Approach for Nonlinear Process MonitoringIEEE Access, 2019
- Sparse Kernel PCA for Outlier Detection2018
- A recursive least square algorithm for online kernel principal component extractionNeurocomputing, 2017
- Sparse kernel principal component analysis based on elastic net regularization2016
- A Survey of Sparse Representation: Algorithms and ApplicationsIEEE Access, 2015
- Compressed sensing signal recovery via forward–backward pursuitDigital Signal Processing, 2013
- Adaptive kernel principal components tracking2012
- A Two-Phase Test Sample Sparse Representation Method for Use With Face RecognitionIEEE Transactions on Circuits and Systems for Video Technology, 2011
- The Accelerated Power Method for Kernel Principal Component AnalysisCommunications in Computer and Information Science, 2011
- Signal Recovery From Random Measurements Via Orthogonal Matching PursuitIEEE Transactions on Information Theory, 2007