Version Of Linear Discriminant Analysis Research Articles

View-Independent Discriminant Analysis with Gradient Self-Similarities for Action Recognition under Multiple Views

In Multi-View Human Action Recognition, the actions are not of single view and hence to achieve an effective recognition performance under multi-view actions, there is a need of multi-view subclass discrimination analysis. Based on this inspiration, this paper proposed a novel action recognition framework based on the Subclass Discriminant Analysis (SDA), an extended version of Linear Discriminant Analysis (LDA). Further, a new key frames selection method is proposed based on Self-Similarity Matrix (SSM), called as Gradient SSM (GSSM). Once the key frames are selected, a composite feature set is extracted through three different set filters such as Gaussian Filter, Gabor filter and Wavelet Filter. Next, the SDA obtain an optimal feature subspace for every action under multiple Views. Finally the SVM algorithm classifies the action. The proposed framework is systematically evaluated on IXMAS dataset and NIXMAS dataset. Experimental results enumerate that our method outperforms the conventional techniques in terms of recognition accuracy.

Prostate lesion delineation from multiparametric magnetic resonance imaging based on locality alignment discriminant analysis.

Multiparametric MRI (mpMRI) has shown promise in the detection and localization of prostate cancer foci. Although techniques have been previously introduced to delineate lesions from mpMRI, these techniques were evaluated in datasets with T2 maps available. The generation of T2 map is not included in the clinical prostate mpMRI consensus guidelines; the acquisition of which requires repeated T2-weighted (T2W) scans and would significantly lengthen the scan time currently required for the clinically recommended acquisition protocol, which includes T2W, diffusion-weighted (DW), and dynamic contrast-enhanced (DCE) imaging. The goal of this study is to develop and evaluate an algorithm that provides pixel-accurate lesion delineation from images acquired based on the clinical protocol. Twenty-five pixel-based features were extracted from the T2-weighted (T2W), apparent diffusion coefficient (ADC), and dynamic contrast-enhanced (DCE) images. The pixel-wise classification was performed on the reduced space generated by locality alignment discriminant analysis (LADA), a version of linear discriminant analysis (LDA) localized to patches in the feature space. Postprocessing procedures, including the removal of isolated points identified and filling of holes inside detected regions, were performed to improve delineation accuracy. The segmentation result was evaluated against the lesions manually delineated by four expert observers according to the Prostate Imaging-Reporting and Data System (PI-RADS) detection guideline. The LADA-based classifier (60±11%) achieved a higher sensitivity than the LDA-based classifier (51±10%), thereby demonstrating, for the first time, that higher classification performance was attained on the reduced space generated by LADA than by LDA. Further sensitivity improvement (75±14%) was obtained after postprocessing, approaching the sensitivities attained by previous mpMRI lesion delineation studies in which nonclinical T2 maps were available. The proposed algorithm delineated lesions accurately and efficiently from images acquired following the clinical protocol. The development of this framework may potentially accelerate the clinical uses of mpMRI in prostate cancer diagnosis and treatment planning.

Spectral regression based marginal Fisher analysis dimensionality reduction algorithm

Abstract Traditional nonlinear dimensionality reduction methods, such as multiple kernel dimensionality reduction and nonlinear spectral regression (SR), are generally regarded as extended versions of linear discriminant analysis (LDA) in the supervised case. As is well known, LDA has the restrictive assumption that the data of each class is of a Gaussian distribution. Thus, the performance of these methods will be degraded if such an assumption is not hold. Although some methods based on marginal Fisher analysis are proposed to overcome the drawback of LDA, they have to solve the problem of dense metrics generalized eigenvalue decomposition, which is very time-consuming. To address these issues, in this paper, marginal Fisher analysis criterion based on extreme learning machine (ELM) is proposed to improve spectral regression and kernel marginal Fisher analysis. It is proved that the proposed marginal Fisher analysis is a special case of traditional kernel marginal Fisher analysis. Based on the proposed criterion, a novel supervised dimensionality reduction algorithm is presented by virtue of ELM and spectral regression. Experimental results on benchmark datasets validate that the proposed algorithm outperforms the state-of-the-art nonlinear dimensionality reduction methods in supervised scenarios.

Predictive Nuclear Chromatin Characteristics of Melanoma and Dysplastic Nevi

Background: The diagnosis of malignant melanoma (MM) is among the diagnostic challenges pathologists encounter on a routine basis. Melanoma may arise in patients with preexisting dysplastic nevi (DN) and it is still the cause of 1.7% of all cancer-related deaths. Melanomas often have overlapping histological features with DN, especially those with severe dysplasia. Nucleotyping for identifying nuclear textural features can analyze nuclear DNA structure and organization. The aim of this study is to differentiate MM and DN using these methodologies. Methods: Dermatopathology slides diagnosed as MM and DN were retrieved. The glass slides were scanned using an Aperio ScanScopeXT at ×40 (0.25 µ/pixel). Whole slide images (WSI) were annotated for nuclei selection. Nuclear features to distinguish between MM and DN based on chromatin distributions were extracted from the WSI. The morphological characteristics for each nucleus were quantified with the optimal transport-based linear embedding in the continuous domain. Label predictions for individual cell nucleus are achieved through a modified version of linear discriminant analysis, coupled with the k-nearest neighbor classifier. Label for an unknown patient was set by the voting strategy with its pertaining cell nuclei. Results: Nucleotyping of 139 patient cases of melanoma (n = 67) and DN (n = 72) showed that our method had superior classification accuracy of 81.29%. This is a 6.4% gain in differentiating MM and DN, compared with numerical feature-based method. The distribution differences in nuclei morphology between MM and DN can be visualized with biological interpretation. Conclusions: Nucleotyping using quantitative and qualitative analyses may provide enough information for differentiating MM from DN using pixel image data. Our method to segment cell nuclei may offer a practical and inexpensive solution in aiding in the accurate diagnosis of melanoma.

Recognition of a Driver's Gaze for Vehicle Headlamp Control

In this paper, we propose a novel method for gaze recognition of a driver coping with rotation of a driver's face. Frontal face images and left half profile images were separately trained using the Viola-Jones (V-J) algorithm to produce classifiers that can detect faces. The right half profile can be detected by mirroring the entire image when neither a frontal face nor a left half profile was detected. As an initial step, this method was used to simultaneously detect the driver's face. Then, we applied a regressional version of linear discriminant analysis (LDAr) to the detected facial region to extract important features for classification. Finally, these features were used to classify the driver's gaze in seven directions. In the feature extraction step, LDAr tries to find features that maximize the ratio of interdistances among samples with large differences in the target value to those with small differences in the target value. Therefore, the resultant features are more fitted to regression problems than conventional feature extraction methods. In addition to LDAr, in this paper, a 2-D extension of LDAr is also developed and used as a feature extraction method for gaze recognition. The experimental results show that the proposed method achieves a good gaze recognition rate under various rotation angles of a driver's head, resulting in a reliable headlamp control performance.

Novel Fisher discriminant classifiers

At the present, several applications need to classify high dimensional points belonging to highly unbalanced classes. Unfortunately, when the training set cardinality is small compared to the data dimensionality (“small sample size” problem) the classification performance of several well-known classifiers strongly decreases. Similarly, the classification accuracy of several discriminative methods decreases when non-linearly separable, and unbalanced, classes are treated. In this paper we firstly survey state of the art methods that employ improved versions of Linear Discriminant Analysis (LDA) to deal with the above mentioned problems; secondly, we propose a family of classifiers based on the Fisher subspace estimation, which efficiently deal with the small sample size problem, non-linearly separable classes, and unbalanced classes. The promising results obtained by the proposed techniques on benchmark datasets and the comparison with state of the art predictors show the efficacy of the proposed techniques.

Ensemble based multi-linear discriminant analysis with boosting and nearest neighbor

The tensor based classifier has attracted a great deal of interest, due to its representation of input objects in a natural form in overcoming small sample size problems and in providing high classification accuracy. Multi-Linear Discriminant Analysis (MLDA) is an efficient classifier, which employs tensor properties to simplify computation and improve accuracy. In this study, a boosting framework is exploited to further improve a tensor-based MLDA classifier. In the boosting framework, several weak learners are trained with different distribution of training samples and, then, integrated with suitable weights to build a strong classifier with a high generalization capacity. In our proposed method, namely BMLDA (Boosted MLDA), the MLDA classifiers are weakened and considered as feature projection components (weak learners) in the boosting framework. Finally, a Nearest Neighbor (NN) classifier makes the final decision and enables the BMLDA to act as a multi-class classifier. To assess BMLDA, several versions of Linear Discriminant Analysis (LDA) classifiers, such as boosted direct LDA, direct LDA, subclass-LDA, MLDA and LDA, were implemented. Empirical evaluations on two real applications demonstrated that the proposed BMLDA outperformed its competitors. The proposed method is beneficial in exploiting the boosting framework to accommodate tensor-based learners that totally construct a powerful multi-class ensemble classifier with much higher performance.

Feature extraction based on subspace methods for regression problems

In this paper, we propose a couple of new feature extraction methods for regression problems. The first one is closely related to the conventional principle component analysis (PCA) but unlike PCA, it incorporates target information in the optimization process and try to find a set of linear transforms that maximizes the distances between points with large differences in target values. On the other hand, the second one is a regressional version of linear discriminant analysis (LDA) which is very popular for classification problems. We have applied the proposed methods to several regression problems and compared the performance with the conventional feature extraction methods. The experimental results show that the proposed methods, especially the extension of LDA, perform well in many regression problems.

Discriminant Locally Linear Embedding With High-Order Tensor Data

Graph-embedding along with its linearization and kernelization provides a general framework that unifies most traditional dimensionality reduction algorithms. From this framework, we propose a new manifold learning technique called discriminant locally linear embedding (DLLE), in which the local geometric properties within each class are preserved according to the locally linear embedding (LLE) criterion, and the separability between different classes is enforced by maximizing margins between point pairs on different classes. To deal with the out-of-sample problem in visual recognition with vector input, the linear version of DLLE, i.e., linearization of DLLE (DLLE/L), is directly proposed through the graph-embedding framework. Moreover, we propose its multilinear version, i.e., tensorization of DLLE, for the out-of-sample problem with high-order tensor input. Based on DLLE, a procedure for gait recognition is described. We conduct comprehensive experiments on both gait and face recognition, and observe that: 1) DLLE along its linearization and tensorization outperforms the related versions of linear discriminant analysis, and DLLE/L demonstrates greater effectiveness than the linearization of LLE; 2) algorithms based on tensor representations are generally superior to linear algorithms when dealing with intrinsically high-order data; and 3) for human gait recognition, DLLE/L generally obtains higher accuracy than state-of-the-art gait recognition algorithms on the standard University of South Florida gait database.

Constructing Modulation Frequency Domain-Based Features for Robust Speech Recognition

Data-driven temporal filtering approaches based on a specific optimization technique have been shown to be capable of enhancing the discrimination and robustness of speech features in speech recognition. The filters in these approaches are often obtained with the statistics of the features in the temporal domain. In this paper, we derive new data-driven temporal filters that employ the statistics of the modulation spectra of the speech features. Three new temporal filtering approaches are proposed and based on constrained versions of linear discriminant analysis (LDA), principal component analysis (PCA), and minimum class distance (MCD), respectively. It is shown that these proposed temporal filters can effectively improve the speech recognition accuracy in various noise-corrupted environments. In experiments conducted on Test Set A of the Aurora-2 noisy digits database, these new temporal filters, together with cepstral mean and variance normalization (CMVN), provide average relative error reduction rates of over 40% and 27% when compared with baseline Mel frequency cepstral coefficient (MFCC) processing and CMVN alone, respectively.

Regularized linear discriminant analysis and its application in microarrays

In this paper, we introduce a modified version of linear discriminant analysis, called the "shrunken centroids regularized discriminant analysis" (SCRDA). This method generalizes the idea of the "nearest shrunken centroids" (NSC) (Tibshirani and others, 2003) into the classical discriminant analysis. The SCRDA method is specially designed for classification problems in high dimension low sample size situations, for example, microarray data. Through both simulated data and real life data, it is shown that this method performs very well in multivariate classification problems, often outperforms the PAM method (using the NSC algorithm) and can be as competitive as the support vector machines classifiers. It is also suitable for feature elimination purpose and can be used as gene selection method. The open source R package for this method (named "rda") is available on CRAN (http://www.r-project.org) for download and testing.

Regularized Linear Discriminant Analysis of Wavelet Compressed Ion Mobility Spectra

Wavelet compression (WC) coupled to a regularized version of linear discriminant analysis (LDA) was evaluated as an effective classifier for identification of benzene, toluene, ethyl benzene, and the xylene isomers (BTEX) from their positive ion mobility spectra. A novel LDA classifier was devised that uses pseudoinverses of the within-group covariance matrices. The pseudoinverse allows LDA to be applied directly to underdetermined data. Wavelet compression improved the classification by removing noise from the spectra. Spectra compressed with the daublet 22 filter yielded better prediction than data compressed by principal components. Ion mobility spectrometry (IMS) affords rapid and sensitive measurements that may be used for process monitoring. This study demonstrates that classification by LDA directly on the wavelet-compressed (WTLDA) positive ion spectra can be used for identification of BTEX compounds. The accompanying benefits include fast model building, removal of noise, and improved classification. Benzene, toluene, ethyl benzene, and o-, m-, and p-xylene at high and low concentrations were used to construct the classification models. Spectra from intermediate concentrations were used for evaluation. Optimization of prediction favored shorter wavelet filters than those obtained from filters optimized for reconstruction.

Penalized discriminant analysis of [15O]-water PET brain images with prediction error selection of smoothness and regularization hyperparameters.

We propose a flexible, comprehensive approach for analysis of [15O]-water positron emission tomography (PET) brain images using a penalized version of linear discriminant analysis (PDA). We applied it to scans from 20 subjects (eight scans/subject) performing a finger movement task and analyzed: 1) two classes to obtain a covariance-normalized baseline-activation image, and 2) eight classes for the mean within subject temporal structure which contained baseline-activation and time-dependent changes in a two-dimensional canonical subspace. We imposed spatial smoothness on the resulting image(s) by expanding it in five tensor-product B-spline (TPS) bases of varying smoothness, and further regularized with a ridge-type penalty on the noise covariance matrix. The discrimination approach of PDA provides a probabilistic framework within which prediction error (PE) estimates are derived. We used these to optimize over TPS bases and a ridge hyperparameter (expressed as equivalent degrees of freedom, EDF). We obtained unbiased, low variance PE estimates using modern resampling tools (.632+ Bootstrap and cross validation), and compared PDA of 1) TPS-projected, mean-normalized and unnormalized scans and 2) mean-normalized scans with and without additional presmoothing. By examining the tradeoffs between PE and EDF, as a function of basis selection and image smoothing we demonstrate the utility of PDA, the PE framework, and the relationship between singular value decomposition and smooth TPS bases in the analysis of functional neuroimages.