Face Data Discriminative Feature Extraction Based on Weighted Maximum Margin Criterion

This paper presents a new method for image feature extraction, namely, the fuzzy 2D discriminant locality preserving projections (F2DDLPP) based on the 2D discriminant locality preserving projections (2DDLPP) and fuzzy set theory. Firstly, we calculate the membership degree matrix by fuzzy k-nearest neighbor (FKNN), then we incorporate the membership degree matrix into the definition of the intra-class scatter matrix and inter-class scatter matrix, respectively. Secondly, we can get the fuzzy intra-class scatter matrix and fuzzy inter-class scatter matrix, respectively. The FKNN is implemented to achieve the distribution information of original samples, and this information is utilized to redefine corresponding scatter matrices. So, F2DDLPP can extract discriminative features from overlapping (outlier) samples which is different to the conventional 2DDLPP. Finally, Experiments on the Yale, ORL face databases, USPS database and PolyU palmprint database are demonstrated to verify the effectiveness of the proposed algorithm.

Traditional hyperspectral image (HSI) classification methods mainly include machine learning and convolutional neural network. However, they extremely depend on the large training samples. To obtain high accuracy on limited training samples, we propose a novel end-to-end spectral-spatial multi-scale network (SSMSN) for HSI classification. The SSMSN uses the multi-scale spectral module and the multi-scale spatial module to extract discriminative multi-scale spectral and multi-scale spatial features separately. In the multi-scale spectral module and spatial module, the multi-scale Res2Net block structure can learn multi-scale features at a granular level and increase the range of receptive fields by constructing the hierarchical residual-like connection within one single residual block. To alleviate the overfitting problem and further improve the classification accuracy on limited training samples, we adopt a simple but effective hinge cross-entropy loss function to train the SSMSN at the dynamic learning rate. A large number of experimental results demonstrate that on the Indiana Pines, University of Pavia, Kennedy Space Center, and Salinas Scene data sets, the proposed SSMSN achieves higher classification accuracy than state-of-the-art methods on limited training samples. Meanwhile, our SSMSN obtains less training and testing time than the popular AUSSC method.

Local graph embedding based on maximum margin criterion via fuzzy set

Linear Discriminant Analysis (LDA) [1] is a well-known method for face recognition in feature extraction and dimension reduction. To solve the “small sample” effect of LDA, Two-Dimensional Linear Discriminant Analysis (2DLDA) [2] has been used for face recognition recently，but its could hardly take use of the relationship between the adjacent scatter matrix. In this paper, I improved the between-class scatter matrix, proposed paired-class scatter matrix for face representation and recognition. In this new method, a paired between-class scatter matrix distance metric is used to measure the distance between random paired between-class scatter matrix. To test this new method, ORL face database is used and the results show that the paired between-class scatter matrix based 2DLDA method （N2DLDA） outperforms the 2DLDA method and achieves higher classification accuracy than the 2DLDA algorithm.

Uncorrelated Local Maximum Margin Criterion: An Efficient Dimensionality Reduction Method for Text Classification

Linear Discriminant Analysis (LDA) is a popular statistical method for both feature extraction and dimensionality reduction in face recognition. The major drawback of LDA is the so-called small sample size (3S) problem. This problem always occurs when the total number of training samples is smaller than the dimension of feature space. Under this situation, the within-class scatter matrix Sw becomes singular and LDA approach cannot be implemented directly. To overcome the 3S problem, this paper proposes a novel wavelet-face based subspace LDA algorithm. Wavelet-face feature extraction and dimensionality reduction are based on two-level D4-filter wavelet transform and discarding the null space of total class scatter matrix St. It is shown that our obtained projection matrix satisfies the uncorrelated constraint conditions. Hence in the sense of statistical uncorrelation, this projection matrix is optimal. The proposed method for face recognition has been evaluated with two public available databases, namely ORL and FERET databases. Comparing with existing LDA-based methods to solve the 3S problem, our method gives the best performance.

In this paper, we introduce an incremental version of recently proposed constrained Linear Discriminant Analysis (LDA). In addition of application in constrained LDA problems, our algorithm which we call Online Discriminative Component Analysis (ODCA) is usable in standard incremental LDA problems. ODCA incrementally computes the solution of LDA with the time complexity lower than most incremental algorithm for LDA while keeps the accuracy of final result as close as possible to offline version. This is done using a special formulation for the scatter matrix updating along with Eigen-space calculation. By exploiting such formulation, the proposed algorithm made capable of updating the solution where a data point added or removed from the problem. It is also usable in problems where its data points have concept drift property. To show efficiency of proposed algorithm, its speed is compared to other existing incremental algorithms as order of complexity. In addition, the classification accuracy of our approach is experimentally compared to other algorithms.

In order to search for an efficient feature extraction approach for face recognition,a method based on feature level fusion was presented.This method integrated Locality Preserving Projection(LPP) with Maximum Margin Criterion(MMC).Firstly,LPP was performed on training samples set,so the projection of each training sample on LPP subspace could be got.Further,MMC algorithm was performed on all the obtained projections to get more efficient discriminant features for recognition.Nearest Neighborhood(NN) algorithm was used to construct classifiers.The experiments on the ORL face database show that the recognition rate of the proposed method is high when pose,illumination condition,face expression and training sample number change.

Locality preserving projections(LPP) algorithm can only preserve nearest local quantity,so within-class and between-class local scatter matrices are introduced for characterizing the manifold structure better,and a method called uncorrelated locality preserving projection analysis based on effective and stable maximum margin criterion(MMC) is proposed.When maximizing the trace difference of scatter matrix,weight the within-class and between-class local scatter matrix through a regularized parameter so as to find the better classification subspace and avoid small sample problem.More importantly,the discriminant feature set based on the MMC is generally statistical correlated,which makes the feature information be redundant,so an uncorrelated constraint was added in the paper and the uncorrelated discriminant feature set is extracted by the derived formulas,which are more favorable for the correct recognition.Ultimately experiments on Yale,PIE face database and MNIST handwritten digit database show that the method in this paper is effective and stable and has a higher correct recognition rate compared with the LPP,LDA(Linear discriminant analysis) and LPMIP(Locality-preserved maximum information projection).

The canonical correlation analysis (CCA) aims at measuring linear relationships between two sets of variables (views) that can be used for feature extraction in classification problems with multiview data. However, the correlated features extracted by the CCA may not be class discriminative, since CCA does not utilize the class labels in its traditional formulation. Although there is a method called discriminative CCA (DCCA) that aims to increase the discriminative ability of CCA inspired from the linear discriminant analysis (LDA), it has been shown that the extracted features with this method are identical to those by the LDA with respect to an orthogonal transformation. Therefore, DCCA is simply equivalent to applying single-view (regular) LDA to each one of the views separately. Besides, DCCA and the other similar DCCA approaches have generalization problems due to the sample covariance matrices used in their computation, which are sensitive to outliers and noisy samples. In this paper, we propose a method, called discriminative alternating regression (D-AR), to explore correlated and also discriminative features. D-AR utilizes two (alternating) multilayer perceptrons, each with a linear hidden layer, learning to predict both the class labels and the outputs of each other. We show that the features found by D-AR on training sets significantly accomplish higher classification accuracies on test sets of facial expression recognition, object recognition, and image retrieval experimental data sets.

Due to a lack of labeled samples, deep learning methods generally tend to have poor classification performance in practical applications. Few-shot learning (FSL), as an emerging learning paradigm, has been widely utilized in hyperspectral image (HSI) classification with limited labeled samples. However, the existing FSL methods generally ignore the domain shift problem in cross-domain scenes and rarely explore the associations between samples in the source and target domain. To tackle the above issues, a graph-based domain adaptation FSL (GDAFSL) method is proposed for HSI classification with limited training samples, which utilizes the graph method to guide the domain adaptation learning process in a uniformed framework. First, a novel deep residual hybrid attention network (DRHAN) is designed to extract discriminative embedded features efficiently for few-shot HSI classification. Then, a graph-based domain adaptation network (GDAN), which combines graph construction with domain adversarial strategy, is proposed to fully explore the domain correlation between source and target embedded features. By utilizing the fully explored domain correlations to guide the domain adaptation process, a domain invariant feature metric space is learned for few-shot HSI classification. Comprehensive experimental results conducted on three public HSI datasets demonstrate that GDAFSL is superior to the state-of-the-art with a small sample size.

Linear discrimination analysis (LDA) technique is an important and well-developed area of image recognition and to date many linear discrimination methods have been put forward. Despite these efforts, there persist in the traditional LDA some weaknesses. In this paper, we propose a new LDA-based method that can overcome the drawback existed in the traditional LDA method. It redefines the between-class scatter by adding a weight function according to the between-class distance, which helps to separate the classes as much as possible. At the same time, in this method, we firstly remove the null space of total scatter matrix which has been proved to be the common null space of both between-class and within-class scatter matrix, and useless for discrimination. Then in the lower-dimensional projected space, the null space of the resulting within-class scatter matrix is calculated. This lower-dimensional null space, combined with the previous projection, represents a subspace of the whole null space of within-class scatter matrix, and is really useful for discrimination. The optimal discriminant vectors of LDA are derived from it. Experiment results show our method achieves better performance in comparison with the traditional LDA methods.

Subspace selection is widely applied in data classification, clustering, and visualization. The samples projected into subspace can be processed efficiently. In this paper, we research the linear discriminant analysis (LDA) and maximum margin criterion (MMC) algorithms intensively and analyze the effects of scatters to subspace selection. Meanwhile, we point out the boundaries of scatters in LDA and MMC algorithms to illustrate the differences and similarities of subspace selection in different circumstances. Besides, the effects of outlier classes on subspace selection are also analyzed. According to the above analysis, we propose a new subspace selection method called angle linear discriminant embedding (ALDE) on the basis of angle measurement. ALDE utilizes the cosine of the angle to get new within-class and between-class scatter matrices and avoids the small sample size problem simultaneously. To deal with high-dimensional data, we extend ALDE to a two-stage ALDE (TS-ALDE). The synthetic data experiments indicate that ALDE can balance the within-class and between-class scatters and be robust to outlier classes. The experimental results based on UCI machine-learning repository and image databases show that TS-ALDE has a lower time complexity than ALDE while processing high-dimensional data.

Face recognition is one of the biometric system that mostly used for individual recognition in the absent machine or access control. This is because the face is the most visible part of human anatomy and serves as the first distinguishing factor of a human being. Feature extraction and classification are the key to face recognition, as they are to any pattern classification task. In this paper, we describe a face recognition method based on Linear Discriminant Analysis (LDA) and k-Nearest Neighbor classifier. LDA used for feature extraction, which directly extracts the proper features from image matrices with the objective of maximizing between-class variations and minimizing within-class variations. The features of a testing image will be compared to the features of database image using K-Nearest Neighbor classifier. The experiments in this paper are performed by using using 66 face images of 22 different people. The experimental result shows that the recognition accuracy is up to 98.33%. Index Terms”face recognition, k nearest neighbor, linear discriminant analysis.

Identification of people by face is the most effective non-intrusive method in biometry. However, it is a great challenge for researchers because faces are complex and multidimensional. In addition, high level of difficulty is added by changes in illumination and/or pose. In this work, a face recognition method based on Higher-Order Statistics (HOS) is proposed. HOS has the important signal processing properties of: (i) handling colored Gaussian measurement noise more efficiently, (ii) extracting information due to deviations from Gaussianity, and (iii) detecting and characterizing nonlinear properties in signals. In this work, features based on HOS are used to build compact signature of faces. It is considered a Public Security scenario where the goal is to detect and identify individuals with criminal links, previously registered in a database. To select the most discriminant HOS-based features, the Fisher's Discriminant Ratio (FDR) was used and the linear correlation was applied to eliminate redundancy. Three classifiers (the Bayesian, Support Vector Machines (SVM) and the K-nearest neighbor (KNN)) were employed for final classification and their performances were compared. For performance evaluation, images from ORL dataset were used. The results showed detection and classification rates over 70% and indicates the potential of HOS on building face signatures.