Kernel Discriminant Research Articles

Use of line-scan Raman hyperspectral imaging to identify corn kernels infected with Aspergillus flavus

The potential of line-scan Raman hyperspectral imaging system equipped with a 785 nm line laser was examined for discrimination of uninfected control, AF36-inoculated and AF13-inoculated corn kernels in this study. The AF13 and AF36 strains were used as representatives of the aflatoxigenic and non-aflatoxigenic A. flavus fungi. A total of 300 kernels were used with 3 treatments, namely, 100 kernels inoculated with the AF13 fungus, 100 kernels inoculated with the AF36 fungus, and 100 kernels inoculated with sterile distilled water as control. The kernels were incubated at 30 °C for 8 days, dried and surface wiped to remove exterior signs of mold. The kernels were imaged on both endosperm and germ sides over the wavenumber range of 103–2831 cm−1. The mean spectrum was extracted from the Raman image of each kernel, and preprocessed with adaptive iteratively reweighted penalized least squares, Savitzky-Golay smoothing and min-max normalization. Based upon the preprocessed group mean spectra, a total of 36 and 51 local Raman peaks were identified from the endosperm side and embryo area of germ side, respectively. With the spectral variables at the identified local peak locations as inputs of discriminant models, the 3-class principal component analysis-linear discriminant analysis models ran 50 random times, achieved mean overall prediction accuracies of 89.47% and 75.55% using the endosperm and embryo data, respectively. The corresponding standard deviations were 3.48% and 4.34%. The results demonstrate usefulness of the line-scan Raman imaging technology in differentiating uninfected control corn kernels and corn kernels infected with aflatoxigenic and non-aflatoxigenic fungi.

A discriminant kernel entropy-based framework for feature representation learning

The intelligent multimedia processing community has developed increasing interest in kernel entropy component analysis (KECA) due to its abilities in effective data transformation and dimensionality reduction. However, since only the unsupervised structural information of Renyi entropy from the given data set is utilized, KECA alone is incapable of generating high quality discriminant features. Aiming to develop a new and generic approach for feature representation learning, this paper proposes a discriminant kernel entropy-based framework, which integrates KECA and a complete discriminant strategy (consisting of regular and irregular discriminant information), to explore discriminant feature representations from the given data set. The framework is realized and further optimized to generate a more powerful discriminant descriptor for feature representation learning, leading to improved performance. Since the joint utilization of kernel entropy estimation and the complete discriminant strategy is able to reveal the distribution and semantic information of the given data, the proposed framework opens up a new front for discriminant feature representation learning via information theoretic learning (ITL). To demonstrate the generic nature and effectiveness of the proposed framework, experiments are conducted on two different data sources; the visual data source (e.g., University of California Irvine (UCI) database, Olivetti Research Lab (ORL) database, Caltech 256 database) and the audio data source (Ryerson Multimedia Lab (RML) audio emotion database). The results show this framework yields superior performance over other methods on the data sets evaluated for feature representation learning.

The classification of free algebras of orthogonal modular forms

We prove a necessary and sufficient condition for the graded algebra of automorphic forms on a symmetric domain of type IV being free. From the necessary condition, we derive a classification result. Let $M$ be an even lattice of signature $(2,n)$ splitting two hyperbolic planes. Suppose $\Gamma$ is a subgroup of the integral orthogonal group of $M$ containing the discriminant kernel. It is proved that there are exactly 26 groups $\Gamma$ such that the space of modular forms for $\Gamma$ is a free algebra. Using the sufficient condition, we recover some well-known results.

HiLAM-aligned kernel discriminant analysis for text-dependent speaker verification

Probabilistic Linear Discriminant Analysis (PLDA) has been a commonly used backend classifier for many text-dependent speaker verification (TDSV) systems. Lately, PLDA projections have been integrated with the traditional Dynamic Time Warping (DTW) template matching framework, resulting in the DTW Online i-vector/PLDA system. The system is shown to achieve state-of-the-art performance for TDSV task. PLDA model serves to train a subspace that compensates for channel and session variabilities. It assumes linear separability between speaker-phrase information and other components. However, this relationship is known to be non-linear. The non-linearity is more prominent in case of short speech extracts, as in the case of the online i-vectors. This results in loss of vital speaker-phrase information at PLDA modeling. To this end, this work explores Kernel Discriminant Analysis (KDA) for TDSV task. It further proposes to use Hierarchical Multi-Layer Acoustic Model (HiLAM) to complement KDA with a more effective speaker-text class definition. The proposed system is hypothesized to benefit on three counts — non-linear modeling ability of KDA, speaker idiosyncrasy information associated with HiLAM-defined speaker-text units and modeling of the exact context of the pass-phrase, as offered by HiLAM. It shows a relative Equal Error Rate (EER) reduction of up to 50.63% on Part 1 of the RSR2015 database when compared to the baseline DTW Online i-vector/PLDA system.

Indoor Localization Algorithm of Terminal Based on RSS Feature Extension and Spectral Regression Kernel Discriminant Analysis

Indoor Localization Algorithm of Terminal Based on RSS Feature Extension and Spectral Regression Kernel Discriminant Analysis

Remote hyperspectral sensing for the early detection of Melanaphis sacchari Zehntner, 1897 (Hemiptera: Aphididae) infestations in sorghum leaves

This paper aims to study the reflectance signature information of infested and non-infested sorghum leaves (Sorghum vulgare L.) by sugarcane aphid (Melanaphis sacchari) to discriminate infested sorghum. The study treatments were 0 (0 aphids/leaf), 1 (1-20 aphids/leaf), 2 (21-50 aphids/leaf), 3 (> = 51 aphids/leaf), 4 (> = 51 aphids/leaf + visible damage), 5 (abiotic stress) and 6 (> = 51 aphids/leaf + abiotic stress). An Ocean OpticsTM HR4000 spectrometer was used. The multifactor ANOVA and Kruskal-Wallis tests at 95% confidence indicated that the reflectance at 402.95, 528.43, 658.36, 788.13, and 965.14 nm wavelengths have significant differences between treatments and with the control. Also Kernel Discriminant analysis was carried out and the combination of the wavelengths centered at 788.17 and 965.14 nm allows 70 % of correct classification of treatments. The results indicate that it is possible to detect M. sacchari infested sorghum by using the spectral information of some specific wavelengths. This study may enable the research of an aerial sensor to make recommendation maps of application pesticides.

Congruence subgroups and orthogonal groups

We derive explicit isomorphisms between certain congruence subgroups of the Siegel modular group, the Hermitian modular group over an arbitrary imaginary-quadratic number field and the modular group over the Hurwitz quaternions of degree 2 and the discriminant kernels of special orthogonal groups SO0(2,n), n=3,4,6. The proof is based on an application of linear algebra adapted to the number theoretical needs.

Class mean vector component and discriminant analysis

• We propose a component analysis method for kernel-based subspace learning. • We give extensive analysis of the method and its connection with KPCA and KDA. • A discriminant analysis method is proposed based on properties of KDA projections. The kernel matrix used in kernel methods encodes all the information required for solving complex nonlinear problems defined on data representations in the input space using simple, but implicitly defined, solutions. Spectral analysis on the kernel matrix defines an explicit nonlinear mapping of the input data representations to a subspace of the kernel space, which can be used for directly applying linear methods. However, the selection of the kernel subspace is crucial for the performance of the proceeding processing steps. We propose a new optimization criterion , leading to a new component analysis method for kernel-based dimensionality reduction that optimally preserves the pair-wise distances of the class means in the feature space. This leads to efficient kernel subspace learning, which is crucial for kernel-based machine learning solutions. We provide extensive analysis on the connections and differences between the proposed criterion and the criteria used in kernel Principal Component Analysis, kernel Entropy Component Analysis and Kernel Discriminant Analysis, leading to a discriminant analysis version of the proposed method. Our theoretical analysis also provides more insights on the properties of the feature spaces obtained by applying these methods. Results on a variety of visual classification problems illustrate the properties of the proposed methods.

Sustainable Development of Polish Macroregions-Study by Means of the Kernel Discriminant Coordinates Method.

The aim of this study was to investigate if the macroregions of Poland are homogeneous in terms of the observed spatio-temporal data characterizing their sustainable development. So far, works related to the sustainable development of selected territorial units have been based on data relating to a specific year rather than many years. The solution to the problem of macroregion homogeneity goes through two stages. In step one, the original spatio-temporal data space (matrix space) was transformed into a kernel discriminant coordinates space. The obtained kernel discriminant coordinates function as synthetic measures of the level of sustainable development of Polish macroregions. These measures contain complete information on the values of 27 diagnostic features examined over 15 years. In the second step, cluster analysis was used in order to identify groups of homogeneous macroregions in the space of kernel discriminant coordinates. The agglomeration method and the Ward method were chosen as commonly used methods. By means of both methods, three super macroregions composed of homogeneous macroregions were identified. Within the kernel discriminant coordinates, the differentiating power of a selected set of 27 features characterizing the sustainable development of macroregions was also assessed. To this end, five different and most commonly used methods of discriminant analysis were used to test the correctness of the classification. Depending on the method, the classification errors amounted to zero or were close to zero, which proves a well-chosen set of diagnostic features. Although the data relate only to a specific country (Poland), the presented statistical methodology is universal and can be applied to any territorial unit and spatial-temporal dynamic data.

A Classification Algorithm by Combination of Feature Decomposition and Kernel Discriminant Analysis (KDA) for Automatic MR Brain Image Classification and AD Diagnosis.

Magnetic resonance (MR) imaging is a widely used imaging modality for detection of brain anatomical variations caused by brain diseases such as Alzheimer's disease (AD) and mild cognitive impairment (MCI). AD considered as an irreversible neurodegenerative disorder with progressive memory impairment moreover cognitive functions, while MCI would be considered as a transitional phase amongst age-related cognitive weakening. Numerous machine learning approaches have been examined, aiming at AD computer-aided diagnosis through employing MR image analysis. Conversely, MR brain image changes could be caused by different effects such as aging and dementia. It is still a challenging difficulty to extract the relevant imaging features and classify the subjects of different groups. This paper would propose an automatic classification technique based on feature decomposition and kernel discriminant analysis (KDA) for classifications of progressive MCI (pMCI) vs. normal control (NC), AD vs. NC, and pMCI vs. stable MCI (sMCI). Feature decomposition would be based on dictionary learning, which is used for separation of class-specific components from the non-class-specific components in the features, while KDA would be applied for mapping original nonlinearly separable feature space to the separable features that are linear. The proposed technique would be evaluated by employing T1-weighted MR brain images from 830 subjects comprising 198 AD patients, 167 pMCI, 236 sMCI, and 229 NC from the Alzheimer's disease neuroimaging initiative (ADNI) dataset. Experimental results demonstrate that classification accuracy (ACC) of 90.41%, 84.29%, and 65.94% can be achieved for classification of AD vs. NC, pMCI vs. NC, and pMCI vs. sMCI, respectively, indicating the promising performance of the proposed method.

Multi-classifier majority voting analyses in provenance studies on iron artefacts

The main objective of this paper is to propose an approach for identification of provenance of archaeological iron artefacts making use of major oxides and trace elements. For this purpose, seven classifiers were built on the basis of the following techniques: Linear Discriminant Analysis (LDA), Support Vector Machine (SVM), Random Forests (RF), Naïve Bayes (NB), K-Nearest Neighbours (KNN), Recursive Partitioning and Regression Trees (RPART) and Kernel Discriminant Analysis (KDA). A final assignment of a given observation to a regional class was carried out on the basis of results provided by all classifiers using a majority voting technique. The proposed approach was first tested on experimental slag and then it was applied to actual archaeological data. It is hoped that this method can become part of a new integrated approach which will consider all available types of data, such as major and trace elements and isotopic ratios.

Chemotyping and identification of protected Dalbergiatimber using gas chromatography quadrupole time of flight mass spectrometry

The international trade in illegally logged and environmentally endangered timber has spurred enforcement agencies to seek additional technical procedures for the identification of wood species. All Dalbergia species are listed under the Convention on International Trade in Endangered Species (CITES) which is the reason this genus was chosen for study. Multiple sources of the heartwood from different Dalbergia species were extracted and chromatographic profiles collected by gas chromatography with high resolution quadrupole Time of Flight mass spectrometry (GC/QToF). The collected data was mined to select peaks and mass ions representative of the investigated Dalbergia species, and used to develop a Microsoft Excel® template offering immediate graphical representation of the results. Using wood specimens sourced from different xylaria, this graphical fingerprint proved adept at definitive identification of Dalbergia species. The CITES Appendix I species, D. nigra, was easily distinguished from D. melanoxylon and look-alike species of other genera. Similarly, a number of other Dalbergia species were differentiated using this current approach. Kernel discrimination analysis (KDA) was applied to increase the confidence of the species identification. The mislabeling of specimens appears to be common, and the emerging technique of GC/QToF in combination with other techniques, offers improved confidence in identification. GC/QToF further provides automation, the dimension of chromatography to avoid interferences, and production of reproducible electron impact positive (EI+) spectra. The prospect of building an EI+ spectral database for future wood identification is an important feature considering the limited accessibility of authenticated wood species specimens.

Kernel sparse representation based model for skin lesions segmentation and classification

Background and ObjectivesMelanoma is a dangerous kind of skin disease with a high death rate, and its prevalence has increased rapidly in recent years. Diagnosis of melanoma in a primary phase can be helpful for its cure. Due to costs for dermatology, we need an automatic system to diagnose melanoma through lesion images. MethodsHere, we propose a sparse representation based method for segmentation and classification of lesion images. The main idea of our framework is based on a kernel sparse representation, which produces discriminative sparse codes to represent features in a high-dimensional feature space. Our novel formulation for discriminative kernel sparse coding jointly learns a kernel-based dictionary and a linear classifier. We also present an adaptive K-SVD algorithm for kernel dictionary and classifier learning. ResultsWe test our approach for both segmentation and classification tasks. The evaluation results on both dermoscopic and digital datasets demonstrate our approach to be competitive as compared to the available state-of-the-art methods, with the advantage of not needing any pre-processing. ConclusionsOur method is insensitive to noise and image conditions and can be used effectively for challenging skin lesions. Our approach is so extensive to be adapted to various medical image segmentations.

Novel Audio Feature Projection Using KDLPCCA-Based Correlation with EEG Features for Favorite Music Classification

A novel audio feature projection using Kernel Discriminative Locality Preserving Canonical Correlation Analysis (KDLPCCA)-based correlation with electroencephalogram (EEG) features for favorite music classification is presented in this paper. The projected audio features reflect individual music preference adaptively since they are calculated by considering correlations with the user's EEG signals during listening to musical pieces that the user likes/dislikes via a novel CCA proposed in this paper. The novel CCA, called KDLPCCA, can consider not only a non-linear correlation but also local properties and discriminative information of each class sample, namely, music likes/dislikes. Specifically, local properties reflect intrinsic data structures of the original audio features, and discriminative information enhances the power of the final classification. Hence, the projected audio features have an optimal correlation with individual music preference reflected in the user's EEG signals, adaptively. If the KDLPCCA-based projection that can transform original audio features into novel audio features is calculated once, our method can extract projected audio features from a new musical piece without newly observing individual EEG signals. Our method therefore has a high level of practicability. Consequently, effective classification of user's favorite musical pieces via a Support Vector Machine (SVM) classifier using the new projected audio features becomes feasible. Experimental results show that our method for favorite music classification using projected audio features via the novel CCA outperforms methods using original audio features, EEG features and even audio features projected by other state-of-the-art CCAs.

Selection of Radiomics Features based on their Reproducibility.

Dimensionality reduction is key to alleviate machine learning artifacts in clinical applications with Small Sample Size (SSS) unbalanced datasets. Existing methods rely on either the probabilistic distribution of training data or the discriminant power of the reduced space, disregarding the impact of repeatability and uncertainty in features.In the present study is proposed the use of reproducibility of radiomics features to select features with high inter-class correlation coefficient (ICC). The reproducibility includes the variability introduced in the image acquisition, like medical scans acquisition parameters and convolution kernels, that affects intensity-based features and tumor annotations made by physicians, that influences morphological descriptors of the lesion.For the reproducibility of radiomics features three studies were conducted on cases collected at Vall Hebron Oncology Institute (VHIO) on responders to oncology treatment. The studies focused on the variability due to the convolution kernel, image acquisition parameters, and the inter-observer lesion identification. The features selected were those features with a ICC higher than 0.7 in the three studies.The selected features based on reproducibility were evaluated for lesion malignancy classification using a different database. Results show better performance compared to several state-of-the-art methods including Principal Component Analysis (PCA), Kernel Discriminant Analysis via QR decomposition (KDAQR), LASSO, and an own built Convolutional Neural Network.

Intrinsic feature extraction using discriminant diffusion mapping analysis for automated tool wear evaluation

We present a method of discriminant diffusion maps analysis (DDMA) for evaluating tool wear during milling processes. As a dimensionality reduction technique, the DDMA method is used to fuse and reduce the original features extracted from both the time and frequency domains, by preserving the diffusion distances within the intrinsic feature space and coupling the features to a discriminant kernel to refine the information from the high-dimensional feature space. The proposed DDMA method consists of three main steps: (1) signal processing and feature extraction; (2) intrinsic dimensionality estimation; (3) feature fusion implementation through feature space mapping with diffusion distance preservation. DDMA has been applied to current signals measured from the spindle in a machine center during a milling experiment to evaluate the tool wear status. Compared with the popular principle component analysis method, DDMA can better preserve the useful intrinsic information related to tool wear status. Thus, two important aspects are highlighted in this study: the benefits of the significantly lower dimension of the intrinsic features that are sensitive to tool wear, and the convenient availability of current signals in most industrial machine centers.

Fusion of PHOG and LDP local descriptors for kernel-based ear biometric recognition

Achieving higher recognition performance in uncontrolled scenarios is a key issue for ear biometric systems. It is almost difficult to generate all discriminative features by using a single feature extraction method. This paper presents an efficient method by combining the two most successful local feature descriptors such as Pyramid Histogram of Oriented Gradients (PHOG) and Local Directional Patterns (LDP) to represent ear images. The PHOG represents spatial shape information and the LDP efficiently encodes local texture information. As the feature sets are curse of high dimension, we used principal component analysis (PCA) to reduce the dimension prior to normalization and fusion. Then, two normalized heterogeneous feature sets are combined to produce single feature vector. Finally, the Kernel Discriminant Analysis (KDA) method is employed to extract nonlinear discriminant features for efficient recognition using a nearest neighbor (NN) classifier. Experiments on three standard datasets IIT Delhi version (I and II) and University of Notre Dame collection E reveal that the proposed method can achieve promising recognition performance in comparison with other existing successful methods.

Single-Trial Detection of Semantic Anomalies From EEG During Listening to Spoken Sentences.

We propose a method for the automatic detection of mismatched feelings that occur in communication. As our first step, we examined the semantically anomalous feelings from EEGs when participants listened to spoken sentences. Previous studies have shown that the event-related potentials (ERP) of an electroencephalogram (EEG) are evoked in the auditory and visual modalities where a semantic anomaly occurs. We expand this knowledge and detect it from a single-trial ERP using machine learning techniques. We recorded the brain activity of eight participants as they listened to sentences that contained semantic anomalies and found that a combination of feature selection using linear discriminant analysis and linear kernel support vector machines achieved the highest accuracy that exceeded 60%. By applying this technique, we plan to detect other types of anomalies in practical situations.

Chemical differentiation of Bolivian Cedrela species as a tool to trace illegal timber trade

Combating illegal timber trade requires the ability to identify species and verify geographic origin of timber. Forensic techniques that independently verify the declared species and geographic origin are needed, as current legality procedures are based on certificates and documents that can be falsified. Timber from the genus Cedrela is among the most economically valued tropical timbers worldwide. Three Cedrela species are included in the Appendix III of CITES: C. fissilis, C. odorata and C. angustifolia (listed as C. lilloi). Cedrela timber is currently traded with false origin declarations and under a different species name, but tools to verify this are lacking. We used Direct Analysis in Real Time Time-of-Flight Mass Spectrometry (DART-TOFMS) to chemically identify Cedrela species and sites of origin. Heartwood samples from six Cedrela species (the three CITES-listed species plus C. balansae, C. montana and C. saltensis) were collected at 11 sites throughout Bolivia. Mass spectra detected by DART-TOFMS comprised 1062 compounds; their relative intensities were analysed using Principal Component Analyses, Kernel Discriminant Analysis (KDA) and Random Forest analyses to check discrimination potential among species and sites. Species were identified with a mean discrimination error of 15–19 per cent, with substantial variation in discrimination accuracy among species. The lowest error was observed in C. fissilis (mean = 4.4 per cent). Site discrimination error was considerably higher: 43–54 per cent for C. fissilis and 42–48 per cent for C. odorata. These results provide good prospects to differentiate C. fissilis from other species, but at present there is no scope to do so for other tested species. Thus, discrimination is highly species specific. Our findings for tests of geographic origin suggest no potential to discriminate at the studied scale and for the studied species. Cross-checking results from different methods (KDA and Random Forest) reduced discrimination errors. In all, the DART-TOFMS technique allows independent verification of claimed identity of certain Cedrela species in timber trade.

Credit scoring analysis using kernel discriminant

Credit scoring model is an important tool for reducing the risk of wrong decisions when granting credit facilities to applicants. This paper investigate the performance of kernel discriminant model in assessing customer credit risk. Kernel discriminant analysis is a non- parametric method which means that it does not require any assumptions about the probability distribution of the input. The main ingredient is a kernel that allows an efficient computation of Fisher discriminant. We use several kernel such as normal, epanechnikov, biweight, and triweight. The models accuracy was compared each other using data from a financial institution in Indonesia. The results show that kernel discriminant can be an alternative method that can be used to determine who is eligible for a credit loan. In the data we use, it shows that a normal kernel is relevant to be selected for credit scoring using kernel discriminant model. Sensitivity and specificity reach to 0.5556 and 0.5488 respectively.