Abstract | ||
---|---|---|
In this work we investigate the problem of gender recognition and develop a novel approach based on Supervised Kernel Principal Components Analysis that demonstrates a significant advantage over more traditional approaches of Linear Discriminant Analysis (LDA), Mixture Discriminant Analysis (MDA), and Support Vector Machines (SVM with RBF-kernel) through 5-fold cross validation. To evaluate the effectiveness of the proposed approach for gender recognition, we use FG-NET Aging database, since it contains faces of very young children as well as senior adults. These two subsets of human faces, young children and senior adults, have been shown by prior researchers to be challenging for gender classification. Both simulation and experiment on FG-NET database suggest that the modified supervised manifold learning approach deconvolves high dimensional features into linearly separable projections that can be easily separated with standard techniques. |
Year | DOI | Venue |
---|---|---|
2015 | 10.1007/978-3-319-23989-7_7 | INTELLIGENCE SCIENCE AND BIG DATA ENGINEERING: IMAGE AND VIDEO DATA ENGINEERING, ISCIDE 2015, PT I |
Field | DocType | Volume |
Kernel (linear algebra),Linear separability,Pattern recognition,Computer science,Support vector machine,Kernel principal component analysis,Artificial intelligence,Linear discriminant analysis,Nonlinear dimensionality reduction,Cross-validation,Principal component analysis | Conference | 9242 |
ISSN | Citations | PageRank |
0302-9743 | 0 | 0.34 |
References | Authors | |
10 | 4 |
Name | Order | Citations | PageRank |
---|---|---|---|
Yishi Wang | 1 | 43 | 5.50 |
Cuixian Chen | 2 | 53 | 6.38 |
Valerie Watkins | 3 | 0 | 0.34 |
Karl Ricanek | 4 | 165 | 18.65 |