PCA versus LDA

By Aleix M. Martínez, A. Kak. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2001

914 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/cadb9a014a4c5bbec57aaf30391f472fa4b69b4d

Citations in the corpus

• Face recognition: component-based versus global approaches (2003, neutral): “A comparison between various state-of-the-art global techniques including eigenfaces, Fisher (cid:1) s discriminant analysis, and kernel PCA can be found in [6,7].”
• Neural Collapse Inspired Feature-Classifier Alignment for Few-Shot Class Incremental Learning (2023, neutral): “It describes an optimal geometric structure for classification due to the minimized within-class variance and the maximized between-class variance (Martinez & Kak, 2001), which indicates that the Fisher Discriminant Ratio (Fisher, 1936; Rao, 1948) is maximized.”
• Detecting Faces in Images: A Survey (2002, neutral): “(See [97] for a discussion about training set size.”
• Eigenfeature Regularization and Extraction in Face Recognition (2008, neutral): “Some works [31], [18] show that when the training data are”
• Multiclass Real-Time Intent Recognition of a Powered Lower Limb Prosthesis (2009, neutral): “As stated by [31], however, PCA may outperform LDA when the inherent distribution of the data is nonnormal.”
• Face recognition using Laplacianfaces (2005, neutral): “Once the Laplacianfaces are created, face recognition [2], [14], [28], [29] becomes a pattern classification task.”
• A Survey of Face Recognition Techniques (2009, neutral): “It should be noted, however, that some recent work [87] shows that when the training data set is small, PCA can outperform LDA and also that PCA is less sensitive to different training sets.”
• Incremental linear discriminant analysis for classification of data streams (2005, neutral): “L INEAR Discriminant Analysis (LDA), also known as Fisher Discriminant Analysis (FDA), seeks directions for efficient discrimination, while another technique known as principle component analysis (PCA) [10], [11] seeks directions efficient for representation.”
• Using discriminant analysis for multi-class classification: an experimental investigation (2006, neutral): “Typical subspace methods are presented in ( Swets & Weng , 1996 ; Martinez & Kak , 2001 ; Yang , et al .”
• Locality Preserving Projections (2003, neutral): “A common way to attempt to resolve this problem is to use dimensionality reduction techniques [7] [9] [29] [48] [52] [66] [70] [73] [83] [87].”