KPCA plus LDA: a complete kernel Fisher discriminant framework for feature extraction and recognition

By Jian Yang, Alejandro F Frangi, Jing-yu Yang, David Dapeng Zhang, Zhong Jin. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2005

899 citations recorded by the source corpus.

This paper examines the theory of kernel Fisher discriminant analysis (KFD) in a Hilbert space and develops a two-phase KFD framework, i.e., kernel principal component analysis (KPCA) plus Fisher linear discriminant analysis (LDA). This framework provides novel insights into the nature of KFD. Based on this framework, the authors propose a complete kernel Fisher discriminant analysis (CKFD) algorithm. CKFD can be used to carry out discriminant analysis in "double discriminant subspaces." The fact that, it can make full use of two kinds of discriminant information, regular and irregular, makes CKFD a more powerful discriminator. The proposed algorithm was tested and evaluated using the FERET face database and the CENPARMI handwritten numeral database. The experimental results show that CKFD outperforms other KFD algorithms.

Subfields: Analytical Chemistry, Computer Vision and Pattern Recognition, Media Technology

View Paper: https://www.semanticscholar.org/paper/4dbc6032886f41d59dc7e112f01a82cfb3149f32

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Analytical Chemistry

In Computer Vision and Pattern Recognition

In Media Technology

Citations in the corpus

Exploiting discriminant information in nonnegative matrix factorization with application to frontal face verification (2006, background): “It is easy to prove, using the theory developed in [6] , that does not contain any discriminant information in respect to the criterion (15) and (16).”
Riemannian Manifold Learning (2008, methodology): “Essentially, dimensionality reduction offers an efficient approach for redundancy removal or data reduction.”
Subclass discriminant analysis (2006, background): “This means that it is not possible to maximize the metric given by B and minimize that of X simultaneously.”
Evolving Unsupervised Deep Neural Networks for Learning Meaningful Representations (2017, background): “Excitingly, Yang et al. [60] have mathematically pointed out that the meaningful representations of the input data lie at its original space.”
Optimal Local Basis: A Reinforcement Learning Approach for Face Recognition (2009, methodology): “2003) and hybrid methods (Yang et al. 2005) are successful examples of applying statistical learning theory to face recognition by introducing a single basis for face representation.”
Face recognition across pose: A review (2009, methodology): “Kernel tricks [36, 54, 66, 80, 82] Nonlinear transformation encapsulated in dimension reduction, simple, fast The existing kernel functions are not specific to pose variations, the choice of kernel functions are limited.”
Enhanced Local Texture Feature Sets for Face Recognition Under Difficult Lighting Conditions (2007, methodology): “Numerous approaches have been proposed, including (among many others) eigenfaces [37,43], ﬁsherfaces [5] and lapla-cianfaces [15], nearest feature line-based subspace analysis [32], neural networks [22,38], elastic bunch graph matching [46], wavelets [46], and kernel methods [48].”
Feature dimensionality reduction: a review (2022, methodology): “Such as kernel-based principal component analysis (KPCA) [69], its basic idea is to transform the input data x into the feature space F through a non-linear mapping (x), and then use the linear PCA on F.”
Eigenfeature Regularization and Extraction in Face Recognition (2008, methodology): “Second, the proposed eigenspectrum regularization not only facilitates this discriminative information search in the whole space but also alleviates the overfitting problem.”
Maximum Neighborhood Margin Discriminant Projection for Classification (2014, methodology): “Among the DR techniques, principal component analysis (PCA) [2, 3], linear discriminant analysis (LDA) [3, 4], locality preserving projection (LPP) [5, 6], and their kernelized and tensorized variants [7–18] are the most representative and well-known algorithms.”
Globally Maximizing, Locally Minimizing: Unsupervised Discriminant Projection with Applications to Face and Palm Biometrics (2007, neutral): “Ç”
Comments on "Globally Maximizing, Locally Minimizing: Unsupervised Discriminant Projection with Application to Face and Palm Biometrics" (2007, methodology): “usingtheglobal-to-localstrategy[ 20 ]tomaketherecognition”