Fisher Linear Discriminant Analysis Research Articles

The Exploring of Student’s School Majoring Data at Madrasah Aliyah Negeri in Samarinda Using Linear Discriminant Analysis Models

Curriculum 2013, which applied to Madrasah Aliyah Negeri (MAN) requires that majors are implemented since the 10th grade The majors is supposed to relate to the study plan when they study at University. The aims of this research are to describe correspondence between current majors and desired future majors, to observe distribution pattern of student exam data based on their boxplot, to classify student majors data using the Linear Discriminant Analysis (LDA) models, and to get the best LDA model to describe the characteristics of majors in MAN. The LDA models used in this research were Fisher’s Linear Discriminant Analysis (FLDA), Diagonal Linear Discriminant Analysis (DLDA), Shrunken Linear Discriminant Analysis (SLDA), Maximum-uncertainty Linear Discriminant Analysis (MLDA), and Factor-model Linear Discriminant Analysis (RFLDA). The best LDA model was chosen based on the classification accuracy produced by resampling with n = 1000 and n = 5000 for training data and testing data with the probability of each data being drawn were 60:40, 70:30, 80:20, and 90:10.

Detection of Music Preferences using Cerebral Blood Flow Signals.

Music has been used for entertainment as well as for medical treatments that remedies physical conditions or psychiatric disorders. It is reported that better outcome can be expected when the music was selected along with personal preferences for these therapies. However it is difficult to find out personalized favorite music if the patient has difficulty in verbal communication due to ageing or developmental disorders. Therefore, this study aimed to develop an objective classification method of music preference based on cortical hemodynamic activities upon listening to the music clips. Fifteen healthy young adults listened their favorite and non-favorite music with their cortical activity measured with 38-channel functional near infrared spectroscopy (fNIRS). Eleven features were extracted from the time-courses of fNIRS signals from the left primary auditory area, the superior temporal gyrus, and the subcentral area. One- to four- dimensional features were individually selected using a leave-one-out cross-validation with two classification algorithms of Fisher linear discriminant analysis (LDA) and support vector machine. The best mean accuracy rate of 92.2 ± 1.7% was obtained when an LDA classifier with four features derived from oxy-hemoglobin signals were adopted showing that our proposed method is valid to classify individual music preference.

Fisher linear discriminant analysis for classification and prediction of genomic susceptibility to stomach and colorectal cancers based on six STR loci in a northern Chinese Han population.

ObjectiveGastrointestinal cancer is the leading cause of cancer-related death worldwide. The aim of this study was to verify whether the genotype of six short tandem repeat (STR) loci including AR, Bat-25, D5S346, ER1, ER2, and FGA is associated with the risk of gastric cancer (GC) and colorectal cancer (CRC) and to develop a model that allows early diagnosis and prediction of inherited genomic susceptibility to GC and CRC.MethodsAlleles of six STR loci were determined using the peripheral blood of six colon cancer patients, five rectal cancer patients, eight GC patients, and 30 healthy controls. Fisher linear discriminant analysis (FDA) was used to establish the discriminant formula to distinguish GC and CRC patients from healthy controls. Leave-one-out cross validation and receiver operating characteristic (ROC) curves were used to validate the accuracy of the formula. The relationship between the STR status and immunohistochemical (IHC) and tumor markers was analyzed using multiple correspondence analysis.ResultsD5S346 was confirmed as a GC- and CRC-related STR locus. For the first time, we established a discriminant formula on the basis of the six STR loci, which was used to estimate the risk coefficient of suffering from GC and CRC. The model was statistically significant (Wilks’ lambda = 0.471, χ2 = 30.488, df = 13, and p = 0.004). The results of leave-one-out cross validation showed that the sensitivity of the formula was 73.7% and the specificity was 76.7%. The area under the ROC curve (AUC) was 0.926, with a sensitivity of 73.7% and a specificity of 93.3%. The STR status was shown to have a certain relationship with the expression of some IHC markers and the level of some tumor markers.ConclusionsThe results of this study complement clinical diagnostic criteria and present markers for early prediction of GC and CRC. This approach will aid in improving risk awareness of susceptible individuals and contribute to reducing the incidence of GC and CRC by prevention and early detection.

Development and Validation of an Esophageal Squamous Cell Carcinoma Detection Model by Large-Scale MicroRNA Profiling

Patients with late-stage esophageal squamous cell carcinoma (ESCC) have a poor prognosis. Noninvasive screening tests using serum microRNAs (miRNAs) to accurately detect early-stage ESCC are needed to improve mortality. To establish a model using serum miRNAs to distinguish patients with ESCC from noncancer controls. In this case-control study, serum miRNA expression profiles of patients with ESCC (n = 566) and control patients without cancer (n = 4965) were retrospectively analyzed to establish a diagnostic model, which was tested in a training set and confirmed in a validation set. Patients histologically diagnosed as having ESCC who did not receive prior therapy or have a past or concurrent cancer other than ESCC were enrolled from the National Cancer Center Hospital in Tokyo, Japan. Between October 2010 and November 2015, control samples were collected from the National Cancer Center Biobank, the Biobank of the National Center for Geriatrics and Gerontology, and the general population undergoing routine health examination. Data analysis was performed between August 2015 and October 2018. Serum samples were randomly divided into discovery and validation sets. The expression of 2565 miRNAs was assessed in each sample. The discriminant model (named the EC index) was evaluated in the training set using Fisher linear discriminant analysis with a greedy algorithm. Receiver operating characteristic curve analysis evaluated the diagnostic ability of the model in the validation set. In the training set, 283 patients with esophageal cancer (median age, 67 years [range, 37-90 years]; 83.4% male) were compared with 283 control patients (median age, 54 years [range, 22-100 years]; 43.1% male), and the EC index was constructed using 6 miRNAs (miR-8073, miR-6820-5p, miR-6794-5p, miR-3196, miR-744-5p, and miR-6799-5p). The area under the receiver operating characteristic curve was 1.00, with sensitivity of 1.00 and specificity of 0.98. The validation set included 283 patients (median age, 66 years [range, 42-87 years]; 83.0% male) and 4682 control patients (median age, 68 years [range, 20-98 years]; 44.7% male), and the area under the receiver operating characteristic curve for the EC index was 1.00, with sensitivity of 0.96 and specificity of 0.98. What appears to be novel serum miRNA discriminant model was developed for the diagnosis of ESCC. A multicenter prospective study is ongoing to confirm the present observations.

Facial Expression Identification System Using fisher linear discriminant analysis and K- Nearest Neighbor Methods.

Facial Expression Identification System Using fisher linear discriminant analysis and K- Nearest Neighbor Methods.

Deep Least Squares Fisher Discriminant Analysis.

While being one of the first and most elegant tools for dimensionality reduction, Fisher linear discriminant analysis (FLDA) is not currently considered among the top methods for feature extraction or classification. In this paper, we will review two recent approaches to FLDA, namely, least squares Fisher discriminant analysis (LSFDA) and regularized kernel FDA (RKFDA) and propose deep FDA (DFDA), a straightforward nonlinear extension of LSFDA that takes advantage of the recent advances on deep neural networks. We will compare the performance of RKFDA and DFDA on a large number of two-class and multiclass problems, many of them involving class-imbalanced data sets and some having quite large sample sizes; we will use, for this, the areas under the receiver operating characteristics (ROCs) curve of the classifiers considered. As we shall see, the classification performance of both methods is often very similar and particularly good on imbalanced problems, but building DFDA models is considerably much faster than doing so for RKFDA, particularly in problems with quite large sample sizes.

Intrusion Detection Model Using Naive Bayes and Deep Learning Technique

The increase of security threats and hacking the computer networks are one of the most dangerous issues should treat in these days. Intrusion Detection Systems (IDSs), are the most appropriate methods to prevent and detect the attacks of networks and computer systems. This study presents several techniques to discover network anomalies using data mining tasks, Machine learning technology and dependence of artificial intelligence techniques. In this research, the smart hybrid model was developed to explore any penetrations inside the network. The model divides into two basic stages. The first stage includes the Genetic Algorithm (GA) in selecting the characteristics with depends on a process of extracting, Discretize And dimensionality reduction through Proportional K-Interval Discretization (PKID) and Fisher Linear Discriminant Analysis (FLDA) on respectively. At the end of the first stage combining Naïve Bayes classifier (NB) and Decision Table (DT) using NSL-KDD data set divided into two separate groups for training and testing. The second stage completely depends on the first stage outputs (predicted class) and reclassified with multilayer perceptrons using Deep Learning4J (DL) and the use of algorithm Stochastic Gradient Descent (SGD). In order to improve the performance in terms of the accuracy in classification of penetrations, raising the average of discovering and reducing the false alarms. The comparison of the proposed model and conventional models show the superiority of the proposed model and the previous conventional hybrid models. The result of the proposed model is 99.9325 of classification accuracy, the rate of detection is 99.9738 and 0.00093 of false alarms

Cross domain association using transfer subspace learning

Transfer learning has gained more attention recently by utilizing knowledge acquired from one domain to advance a learning performance in another domain. Existing homogeneous transfer learning methods have progressed to a point where feature spaces are common in training and testing domains. However, heterogeneous transfer learning is still in its nascent stage where features of training and testing domains are different. Taking this into account, Bregman Divergence Regularization is used to minimize the probability distribution difference between training and testing domains and to take them together to a shared subspace. To discriminate data within individual domains, a projection matrix is obtained using Fisher Linear Discriminant Analysis subspace learning algorithm. Experimentation is performed on two efficiently used biometrics: the face and fingerprint. Two types of cross-domain settings are used: (1) Face + Finger2Finger where training samples come from face (labeled samples) and fingerprint (unlabeled samples) data sets, while testing is performed on a fingerprint dataset. (2) Finger + Face2Face where training samples come from fingerprint (labeled samples) and face (unlabeled samples) data sets while testing is performed on a face dataset. This paper proposes a cross domain association between face and fingerprint that finds utility in forensic applications.

Classification of Alzheimer disease among susceptible brain regions

AbstractStatistical and machine learning techniques are frequently employed in the study of neuroimaging data for finding Alzheimer disease (AD) in clinical studies and in additional inquiries about research settings. AD affects the whole brain and as a result the quality of life, where most affected regions are the hippocampus (HP), middle temporal gyrus (MTG), entorhinal cortex, and posterior cingulate cortex (PCC). We used well‐known classification methods to diagnose the affected regions of the brain at different stages of age using biomarker modalities and functional magnetic resonance imaging (fMRI) at the resting state, and later marked the affected brain region on MRI. We have used well‐known support vector machine (SVM), Fisher's linear discriminant analysis, artificial neural network, and logistic regression for the classification of AD. In the context of receiver operating characteristic (ROC) curves, an SVM provided the best classification among AD stages. Moreover, analysis showed development of AD.

Parametric Fault Diagnosis of Analog Circuits Based on a Semi-Supervised Algorithm

The parametric fault diagnosis of analog circuits is very crucial for condition-based maintenance (CBM) in prognosis and health management. In order to improve the diagnostic rate of parametric faults in engineering applications, a semi-supervised machine learning algorithm was used to classify the parametric fault. A lifting wavelet transform was used to extract fault features, a local preserving mapping algorithm was adopted to optimize the Fisher linear discriminant analysis, and a semi-supervised cooperative training algorithm was utilized for fault classification. In the proposed method, the fault values were randomly selected as training samples in a range of parametric fault intervals, for both optimizing the generalization of the model and improving the fault diagnosis rate. Furthermore, after semi-supervised dimensionality reduction and semi-supervised classification were applied, the diagnosis rate was slightly higher than the existing training model by fixing the value of the analyzed component.

Efficient Segmentation of Brain Tumor Using FL-SNM with a Metaheuristic Approach to Optimization.

Nowadays, automatic tumor detection from brain images is extremely significant for many diagnostic as well as therapeutic purposes, due to the unpredictable shape and appearance of tumors. In medical image analysis, the automatic segmentation of tumors from brain using magnetic resonance imaging (MRI) data is the most critical issue. Existing research has some limitations, such as high processing time and lower accuracy, because of the time required for the training process. In this research, a new automatic segmentation process is introduced using machine learning and a swarm intelligence scheme. Here, a fuzzy logic with spiking neuron model (FL-SNM) is proposed for segmenting the brain tumor region in MR images. Initially, input images are preprocessed to remove Gaussian and Poisson noise using a modified Kuan filter (MKF). In the MKF, the optimal selection of the minimum MSE of image pixels is achieved using a random search algorithm (RSA), which improves the peak signal-to-noise ratio (PSNR). Then, the image is smoothed using an anisotropic diffusion filter (ADF) to reduce the over-filtering problem. Afterwards, to extract statistical texture features, Fisher's linear-discriminant analysis (FLDA) is used. Finally, extracted features are transferred to the FL-SNM process and this scheme effectively segments the tumor region. In FL-SNM, the consequent parameters such as weight and bias play an important role in segmenting the region. Therefore, optimizing the weight parameter values using a chicken behavior-based swarm intelligence (CSI) algorithm, is proposed. The proposed (FL-SNM) scheme attained better performance in terms of high accuracy (94.87%), sensitivity (92.07%), specificity (99.34%), precision rate (89.36%), recall rate (88.39%), F-measure (95.06%), G-mean (95.63%), and DSC rate (91.2%), compared to existing convolutional neural networks (CNNs) and hierarchical self-organizing maps (HSOMs).

A Multi-Feature LED Bit Detection Algorithm in Vehicular Optical Camera Communication

In a vehicular optical camera communication (VOCC) system, digital information is transmitted using LED panels and received using cameras. The transmitted bits are obtained by processing the captured images to detect the ON and OFF statuses of LEDs in the array. In determining the LED status, the current LED bit detection algorithms only rely on the grayscale, which is an unreliable feature of LEDs. Consequently, they exhibit poor performance in unfavorable conditions. The contribution of this paper is the proposed multi-feature LED bit detection algorithm that employs three new features of LED: average greyscale ratio (AGR), gradient radial inwardness (GRI), and neighbor greyscale ratio (NGR). Two features, AGR and GRI, individually have substantially more discriminability of LED statuses than greyscale. More importantly, the three proposed features differentiate LED statuses under different perspectives. Consequently, the combination of the three features using Fisher linear discriminant analysis (FLDA) yields outstanding accuracy and robustness of bit detection, even in severe conditions. Highly realistic simulations of a VOCC system are conducted to verify the superiority and robustness of the proposed algorithm.

The failure prediction models: a comparative study

The objective of this paper is to explain and predict the companies' failure risk of claiming a credit in order to improve the decision-making process. Our tests are conducted on a sample of 513 French companies and use the methodologies of principal component analysis, Fisher's linear discriminant analysis; and logistic regression. All have been implemented using carefully selected economic and financial ratios. The empirical results show that the turnover in logarithm, financial independence, the by-employee turnover and the turnover growth rate are the most important explanatory variables with a ranking success that exceeds 70%. This finding is consistent across all tested models and specifications.

An efficient approach for handling degradation in character recognition

Recognition of historical printed degraded Kannada characters is not solved completely and remains as a challenge to the researchers still. In this paper, a scale for measuring degradation of a character is proposed. Further, the degradation is characterised to high, medium and low based on this scale, and use it to study the efficiency of the character restoration technique designed. A new approach, fit discriminant analysis (FDA) for recognition is proposed and compares its recognition accuracy with the existing techniques support vector machines (SVM) and Fisher linear discriminant (FLD) analysis. Through extensive experimentation it is established that rebuilding of characters improves the recognition accuracy of learning-based approaches SVM, FDA, and FLD significantly. Further, it is established that the proposed approach FDA gives the best recognition accuracy for historical printed degraded documents. It is also proved that training-testing set applying the proposed degradation measure is required for better recognition accuracy.

An efficient approach for handling degradation in character recognition

Recognition of historical printed degraded Kannada characters is not solved completely and remains as a challenge to the researchers still. In this paper, a scale for measuring degradation of a character is proposed. Further, the degradation is characterised to high, medium and low based on this scale, and use it to study the efficiency of the character restoration technique designed. A new approach, fit discriminant analysis (FDA) for recognition is proposed and compares its recognition accuracy with the existing techniques support vector machines (SVM) and Fisher linear discriminant (FLD) analysis. Through extensive experimentation it is established that rebuilding of characters improves the recognition accuracy of learning-based approaches SVM, FDA, and FLD significantly. Further, it is established that the proposed approach FDA gives the best recognition accuracy for historical printed degraded documents. It is also proved that training-testing set applying the proposed degradation measure is required for better recognition accuracy.

FPCANet: Fisher discrimination for Principal Component Analysis Network

With the development of Deep Learning (DL) in recent years, integrating traditional machine learning methods with DL has received a lot of attention. One of such representative work is the Principal Component Analysis Network (PCANet), which adopts Principal Component Analysis (PCA) to learn convolutional kernels (or filters) for image classification. Nevertheless, PCANet does not use the discriminative information during learning filters. In this paper, based on PCA in the PCANet, we propose a new model called Fisher PCA (FPCA) which combines Fisher Linear Discriminant Analysis (LDA) with PCA. To facilitate the practical calculation, a approximate model of FPCA is given by introducing a intermediate variable. Theoretically, we analyze the relationship between the original FPCA model and its approximate model, and give a convergence analysis of the approximate model. Additionally, stacking the approximate model of FPCA, we also construct a deep network named FPCA Network (FPCANet). Extensive experiments are conducted to compare FPCANet with other state-of-the-art models for classification problems. The results show that the proposed FPCANet can learn features with more discriminative information, and thus demonstrating its competitive performances on classification tasks.

Multiple Cytokine Profiling: A New Model to Predict Response to Tumor Necrosis Factor Antagonists in Ulcerative Colitis Patients.

Ulcerative colitis (UC) is a form of inflammatory bowel disease, and antibodies against tumor necrosis factor (anti-TNF) are used for treatment. Many patients are refractory or lose response to anti-TNF, and predicting response would be an extremely valuable clinical tool. Unlike most biomarkers, cytokines directly mediate inflammation, and their measurement may predict the likelihood of response or no response. Serum samples were obtained from 49 UC patients before infliximab infusions, and levels of 17 cytokines were measured using a multiplex assay. The Fisher linear discriminant analysis (FLDA) was applied to the cytokine values to predict which patients would respond to infliximab. "Response" was defined as clinical remission after the third infusion, and "no response" was defined as lack of remission after the third infusion. The Fisher linear discriminant analysis model identified a subset of seven predictor cytokines: TNF-α, IL-12, IL-8, IL-2, IL-5, IL1-β, and IFN-γ. The obtained canonical coefficients enabled to calculate discriminant scores as linear combinations of the cytokines; model classified thepatients as responders and nonresponders with a sensitivity of 84.2% and a specificity of 93.3%. Overall, the yield of the FLDA model was 89.8% of the total 49 patients. An unbiased, statistically derived, predictive model based on measurement of serum cytokines before therapy may predict a positive or negative outcome from the administration of anti-TNF to UC patients. Because accurately measuring cytokines is simple and inexpensive, the model may be a valuable new tool to complement other laboratory parameters used in the management of IBD patients.

Adulterant identification in mutton by electronic nose and gas chromatography-mass spectrometer

Battling meat adulteration is essential in preserving both public health and a fair market. This study depicts a method of reducing adulteration in meats, in which Electronic-nose (E-nose) and gas chromatography-mass spectrometer (GC-MS) were applied to identify adulterants in mutton. Duck meat was selected as a model adulterant due to its lower cost and similarity in flavor to mutton, as well as being frequently adulterated in China. Qualitative and quantitative analysis were conducted using linear regression, fisher linear discriminant analysis (FLDA), and multilayer perceptron neural networks analysis (MLPN) on E-nose signals. Several fingerprint volatile organic chemicals (VOC) were identified by GC-MS to validate the E-nose results. Multivariate partial least square regression (PLS) was carried out to study the relationships between GC-MS and E-nose. The results of GC-MS confirmed that E-nose can be used to identify duck adulteration in mutton, with a minimum detection ratio of 10%. This method proved that rapid detection of mutton adulterated duck meat using E-nose has a high accuracy, which has reduced detecting time and improved detection efficiency.

Discriminating growth stages of an endangered Mediterranean relict plant (Ammopiptanthus mongolicus) in the arid Northwest China using hyperspectral measurements

Ammopiptanthus mongolicus, the only drought-resistant, leguminous, evergreen shrub in the desert region of China, is endangered due to climate change and its growth stages urgently need to be non-destructively detected. Although many spectral indexes have been proposed for characterizing vegetation, the relationships are often inconsistent, making it challenging to characterize the status of vegetation across all growth stages. This study investigated the Spectral Features of the endangered desert plant A. mongolicus at different growth stages, and extracted the identified Spectral Features for the establishment of detection and discrimination models using Partial Least Square Regression (PLSR) and Fisher Linear Discriminate Analysis (FLDA), respectively. The results showed spectral reflectance of A. mongolicus differed across different growth stages and it generally increased with the degree of senescence. Poor performance was found in the single factor model, with RMSE ranging from 20.34 to 27.39 or Overall Accuracy of 60% in the validation datasets. The multivariate PLSR model, based on Normalized Difference Vegetation Index (NDVI), Normalized Difference Red Edge Index (NDRE), Physiological Reflectance Index (PhRI) and Plant Senescence Reflectance Index (PSRI), turned out to be accurate in detecting the growth stages, with R2 of 0.89 and RMSE of 12.46, and the performance of the multivariate FLDA model based on 14 Spectral Features was acceptable, with an Overall Accuracy of 89% in the validation datasets. This research provides useful insights for timely and non-destructively discriminating different growth stages by using multivariate PLSR and FLDA analysis.

Spectral Characterization of Eight Marine Phytoplankton Phyla and Assessing a Pigment-Based Taxonomic Discriminant Analysis for the in Situ Classification of Phytoplankton Blooms.

Early stage identification of harmful algal blooms (HABs) has gained significance for marine monitoring systems over the years. Various approaches for in situ classification have been developed. Among them, pigment-based taxonomic classification is one promising technique for in situ characterization of bloom compositions, although it is yet underutilized in marine monitoring programs. To demonstrate the applicability and importance of this powerful approach for monitoring programs, we combined an ultra low-cost and miniaturized multichannel fluorometer with Fisher's linear discriminant analysis (LDA). This enables the real-time characterization of algal blooms at order level based on their spectral properties. The classification capability of the algorithm was examined with a leave-one-out cross validation of 53 different unialgal cultures conducted in terms of standard statistical measures and independent figures of merit. The separation capability of the linear discriminant analysis was further successfully examined in mixed algal suspensions. Besides this, the impact of the growing status on the classification capability was assessed. Further, we provide a comprehensive study of spectral features of eight different phytoplankton phyla including an extensive study of fluorescence excitation spectra and marker pigments analyzed via HPLC. The analyzed phytoplankton species belong to the phyla of Cyanobacteria, Dinophyta (Dinoflagellates), Bacillariophyta (Diatoms), Haptophyta, Chlorophyta, Ochrophyta, Cryptophyta, and Euglenophyta.