Popular Feature Extraction Methods Research Articles

A Deep Learning Method for Bearing Cross-Domain Fault Diagnostics Based on the Standard Envelope Spectrum.

Intelligent fault diagnostics based on deep learning provides a favorable guarantee for the reliable operation of equipment, but a trained deep learning model generally has low prediction accuracy in cross-domain diagnostics. To solve this problem, a deep learning fault diagnosis method based on the reconstructed envelope spectrum is proposed to improve the ability of rolling bearing cross-domain fault diagnostics in this paper. First, based on the envelope spectrum morphology of rolling bearing failures, a standard envelope spectrum is constructed that reveals the unique characteristics of different bearing health states and eliminates the differences between domains due to different bearing speeds and bearing models. Then, a fault diagnosis model was constructed using a convolutional neural network to learn features and complete fault classification. Finally, using two publicly available bearing data sets and one bearing data set obtained by self-experimentation, the proposed method is applied to the data of the fault diagnostics of rolling bearings under different rotational speeds and different bearing types. The experimental results show that, compared with some popular feature extraction methods, the proposed method can achieve high diagnostic accuracy with data at different rotational speeds and different bearing types, and it is an effective method for solving the problem with cross-domain fault diagnostics for rolling bearings.

Can using a pre-trained deep learning model as the feature extractor in the bag-of-deep-visual-words model always improve image classification accuracy?

This article investigates whether higher classification accuracy can always be achieved by utilizing a pre-trained deep learning model as the feature extractor in the Bag-of-Deep-Visual-Words (BoDVW) classification model, as opposed to directly using the new classification layer of the pre-trained model for classification. Considering the multiple factors related to the feature extractor -such as model architecture, fine-tuning strategy, number of training samples, feature extraction method, and feature encoding method-we investigate these factors through experiments and then provide detailed answers to the question. In our experiments, we use five feature encoding methods: hard-voting, soft-voting, locally constrained linear coding, super vector coding, and fisher vector (FV). We also employ two popular feature extraction methods: one (denoted as Ext-DFs(CP)) uses a convolutional or non-global pooling layer, and another (denoted as Ext-DFs(FC)) uses a fully-connected or global pooling layer. Three pre-trained models-VGGNet-16, ResNext-50(32×4d), and Swin-B-are utilized as feature extractors. Experimental results on six datasets (15-Scenes, TF-Flowers, MIT Indoor-67, COVID-19 CXR, NWPU-RESISC45, and Caltech-101) reveal that compared to using the pre-trained model with only the new classification layer re-trained for classification, employing it as the feature extractor in the BoDVW model improves the accuracy in 35 out of 36 experiments when using FV. With Ext-DFs(CP), the accuracy increases by 0.13% to 8.43% (averaged at 3.11%), and with Ext-DFs(FC), it increases by 1.06% to 14.63% (averaged at 5.66%). Furthermore, when all layers of the pre-trained model are fine-tuned and used as the feature extractor, the results vary depending on the methods used. If FV and Ext-DFs(FC) are used, the accuracy increases by 0.21% to 5.65% (averaged at 1.58%) in 14 out of 18 experiments. Our results suggest that while using a pre-trained deep learning model as the feature extractor does not always improve classification accuracy, it holds great potential as an accuracy improvement technique.

Sparse discriminant PCA based on contrastive learning and class-specificity distribution

Much mathematical effort has been devoted to developing Principal Component Analysis (PCA), which is the most popular feature extraction method. To suppress the negative effect of noise on PCA performance, there have been extensive studies and applications of a large number of robust PCAs achieving outstanding results. However, existing methods suffer from at least two shortcomings: (1) They expressed PCA as a reconstruction model measured by Euclidean distance, which only considers the relationship between the data and its reconstruction and ignores the differences between different data points; (2) They did not consider the class-specificity distribution information contained in the data itself, thus lacking discriminative properties. To overcome the above problems, we propose a Sparse Discriminant Principal Components Analysis (SDPCA) model based on contrastive learning and class-specificity distribution. Specifically, we use contrastive learning to measure the relationship between samples and their reconstructions, which fully takes the discriminative information between data into account in PCA. In order to make the extracted low-dimensional features profoundly reflect the class-specificity distribution of the data, we minimize the squared ℓ1,2-norm of the low-dimensional embedding. In addition, to reduce the effects of redundant features and noise and to improve the interpretability of PCA at the same time, we impose sparsity constraints on the projection matrix using the squared ℓ1,2-norm. Our experimental results on different types of benchmark databases demonstrate that our model has state-of-the-art performance.

Performance Analysis of Feature Extraction Approach: Local Binary Pattern and Principal Component Analysis for Iris Recognition system

Many techniques have been proposed for the recognition of Iris. Most of them are single resolution techniques which results in poor performance. In this paper, feature extraction approaches like local binary pattern and principal component analysis assimilation has been offered. For classification, Support Vector Machine has been used. This paper compares the efficiency of two popular feature extraction methods Principal Component Analysis and Local Binary Pattern using two different iris databases CASIA and UBIRIS. The models were tested using 200 iris images. Statistical parameters like F1 score and Accuracy are tested for different threshold values. Our proposed method results with accuracy of 94 and 92%, is obtained for using Local Binary Pattern for CASIA and UBIRIS data set respectively. The Receiver Operating Characteristic Curve has been drawn and Area under Curve is also calculated. The experiment has been extended by varying the dataset sizes. The result shows that LBP achieves better performance with both CASIA and UBIRIS databases compared to PCA.

Spectral segmentation based dimension reduction for hyperspectral image classification

ABSTRACT Hyperspectral images (HSI) contain a wide range of information, the most prominent technology for observing the earth. However, using an original HSI high-dimensional datacube, the classification task faces significant challenges since it has a high computational cost. As a result, dimensionality reduction is indispensable. A dimension reduction method has been introduced in this paper, including feature extraction and feature selection to obtain feature subsets. Minimum Noise Fraction (MNF) is a popular feature extraction method for HSI, requiring a high computational capability. We propose a segmented MNF that divides the complete HSI into groups utilising normalised cross-cumulative residual entropy (nCCRE). An nCCRE-based feature selection is also employed to improve the quality of the chosen features using the max-relevancy min-redundancy measure. The support vector machine (SVM) classifier is used on two real HSI to evaluate the efficiency of the extracted subsets.

The Effect of the MFCC Frame Length in Automatic Voice Pathology Detection

Automatic voice pathology detection is a research topic, which has gained increasing interest recently. Although methods based on deep learning are becoming popular, the classical pipeline systems based on a two-stage architecture consisting of a feature extraction stage and a classifier stage are still widely used. In these classical detection systems, frame-wise computation of mel-frequency cepstral coefficients (MFCCs) is the most popular feature extraction method. However, no systematic study has been conducted to investigate the effect of the MFCC frame length on automatic voice pathology detection. In this work, we studied the effect of the MFCC frame length in voice pathology detection using three disorders (hyperkinetic dysphonia, hypokinetic dysphonia and reflux laryngitis) from the Saarbrücken Voice Disorders (SVD) database. The detection performance was compared between speaker-dependent and speaker-independent scenarios as well as between speaking task -dependent and speaking task -independent scenarios. The Support Vector Machine, which is the most widely used classifier in the study area, was used as the classifier. The results show that the detection accuracy depended on the MFFC frame length in all the scenarios studied. The best detection accuracy was obtained by using a MFFC frame length of 500 ms with a shift of 5 ms.

Finger Vein Authentication Based on Wavelet Scattering Networks

Biometric-based authentication systems have attracted more attention than traditional authentication techniques such as passwords in the last two decades. Multiple biometrics such as fingerprint, palm, iris, palm vein and finger vein and other biometrics have been introduced. One of the challenges in biometrics is physical injury. Biometric of finger vein is of the biometrics least exposed to physical damage. Numerous methods have been proposed for authentication with the help of this biometric that suffer from weaknesses such as high computational complexity and low identification rate. This paper presents a novel method of scattering wavelet-based identity identification. Scattering wavelet extracts image features from Gabor wavelet filters in a structure similar to convolutional neural networks. What distinguishes this algorithm from other popular feature extraction methods such as deep learning methods, filter-based methods, statistical methods, etc., is that this algorithm has very high skill and accuracy in differentiating similar images but belongs to different classes, even when the image is subject to serious damage such as noise, angle changes or pixel location, this descriptor still generates feature vectors in a way that minimizes classifier error. This improves classification and authentication. The proposed method has been evaluated using two databases Finger Vein USM (FV-USM) and Homologous Multi-modal biometrics Traits (SDUMLA-HMT). In addition to having reasonable computational complexity, it has recorded excellent identification rates in noise, rotation, and transmission challenges. At best, it has a 98.2% identification rate for the SDUMLA-HMT database and a 96.1% identification rate for the FV-USM database.

A combination of feature extraction methods and deep learning for brain tumour classification

This paper presents a method for categorizing tumour disease from magnetic resonance imaging images using a convolutional neural network. The proposed technique consists of three major phases, including feature extraction, feature selection, and combination. The authors considered the classification method using convolutional neural network without any pre-processing on input images as the original method. The original method is then improved in some sequential phases when convolutional neural network uses features generated by feature extraction methods. Many popular feature extraction methods in generating the input features of the network are examined. After examining the results, a set of feature extraction schemes with appropriate performance are selected for the next phase. Also, the authors assigned weight factors to each of the selected methods, according to their accuracy. By assigning these weight factors to the methods, the network's accuracy increased to an acceptable level compared to the first and second observations. An accuracy above 99.76% was achieved, which is a 2% improvement using these feature extraction methods compared to the original method. The authors also added a third class named ‘I do not know’ to increase the tumour detection problem's reliability in the last phase. The authors have successfully introduced a convolutional neural network architecture with high accuracy and reliability.

Probabilistic Models for Common Spatial Patterns: Parameter-Expanded EM and Variational Bayes

Common spatial patterns (CSP) is a popular feature extraction method for discriminating between positive andnegative classes in electroencephalography (EEG) data.Two probabilistic models for CSP were recently developed: probabilistic CSP (PCSP), which is trained by expectation maximization (EM), and variational BayesianCSP (VBCSP) which is learned by variational approx-imation. Parameter expansion methods use auxiliaryparameters to speed up the convergence of EM or thedeterministic approximation of the target distributionin variational inference. In this paper, we describethe development of parameter-expanded algorithms forPCSP and VBCSP, leading to PCSP-PX and VBCSP-PX, whose convergence speed-up and high performanceare emphasized. The convergence speed-up in PCSP-PX and VBCSP-PX is a direct consequence of parame-ter expansion methods. The contribution of this study is the performance improvement in the case of CSP,which is a novel development. Numerical experimentson the BCI competition datasets, III IV a and IV 2ademonstrate the high performance and fast convergenceof PCSP-PX and VBCSP-PX, as compared to PCSP andVBCSP.

Fake News Classification Using Random Forest and Decision Tree (J48)

FakeNews is one of the most popular phenomena that have considerable effects on our social life, especially in the political domain. Nowadays, creating fake news becomes very easy because of users' widespread using the internet and social media. Therefore, the detection of elusiveness news is a crucial problem that needs to be considerable mainly because of its challenges like the limited amount of the benchmark datasets and the amount of the published news every second. This research proposed utilizing two different machine learning algorithms (random forest and decision tree (J48)) to detect the fake news. In this paper, the full dataset size equals 20,761 samples, while the testing sample size equals 4,345 samples.The preprocessing steps start with cleaning data by removing unnecessary special characters, numbers, English letters, and white spaces, and finally, removing stop words is implemented. After that, the most popular feature extraction method (TF-IDF) is used before applying the two suggested classification algorithms. The results show that the best accuracy achieved equals 89.11% using the decision tree model while using the random forest; the accuracy achieved equals 84.97 %.

Performances of dimension reduction techniques for welding quality prediction based on the dynamic resistance signal

Abstract In this investigation, the welding quality and the dynamic resistance signal in the welding process were correlated using different dimension reduction techniques and regression models. The 0.4-mm-thickness TC2 titanium alloy was used as the welding material, while the welding experiments were carried out by a small pneumatic high-frequency alternating current welding machine. The Rogowski coil and alligator type wire clips were respectively utilized to collect the welding current and voltage in the welding process. The mechanism of dynamic resistance signal changes was studied and explained. In order to extract the features related to the welding quality and reveal the efficiency of the automatic feature extraction method based on the dimension reduction techniques, several dynamic resistance signals with different welding process parameters were processed by the approaches of principal component analysis (PCA), isometric mapping (Isomap), and locally linear embedding (LLE). After that, the redundant information in the dynamic resistance signal which has little to do with the welding quality would be removed and the useful features could also be isolated. And then three regression models quantifying the extracted features and the nugget size were created and their performances were also compared with each other. The results implied that the regression model based on the LLE technique was more robust than those established on the basis of the features automatically extracted from the dynamic resistance signal using PCA and Isomap. Compared with the existing popular manual feature extraction methods in the previous research work, the method proposed in this investigation outperforms the other methods in terms of two aspects. It not only can minimize the prior assumptions about the certain shape of the dynamic resistance curve and remove the subjective factors caused by the manual extraction method, but can assess and monitor the welding quality with a good level of reliability.

A Computationally Efficient Multiclass Time-Frequency Common Spatial Pattern Analysis on EEG Motor Imagery.

Common spatial pattern (CSP) is a popular feature extraction method for electroencephalogram (EEG) motor imagery (MI). This study modifies the conventional CSP algorithm to improve the multi-class MI classification accuracy and ensure the computation process is efficient. The EEG MI data is gathered from the Brain-Computer Interface (BCI) Competition IV. At first, a bandpass filter and a timefrequency analysis are performed for each experiment trial. Then, the optimal EEG signals for every experiment trials are selected based on the signal energy for CSP feature extraction. In the end, the extracted features are classified by three classifiers, linear discriminant analysis (LDA), naïve Bayes (NVB), and support vector machine (SVM), in parallel for classification accuracy comparison.The experiment results show the proposed algorithm average computation time is 37.22% less than the FBCSP (1st winner in the BCI Competition IV) and 4.98% longer than the conventional CSP method. For the classification rate, the proposed algorithm kappa value achieved 2nd highest compared with the top 3 winners in BCI Competition IV.

Common Cross-Spectral Patterns of Electroencephalography for Reliable Cognitive Task Identification

A current popular feature extraction method of classifying cognitive states and task engagements from electroencephalographs (EEG) is common spatial patterns (CSP). However, the classical CSP only focuses on the correlation between the signals and ignores all characteristics of the signals in the time domain and the frequency domain. In this paper we propose Common cross-spectral patterns (CCSP) a novel EEG feature extraction method for combining spectral and spatial patterns based on cross-spectral density (CSD) to overcome the disadvantages of classical CSP. In CCSP method. the cross-power matrices (CPMs) are extracted to measure the spatial correlation of each task in the band of interest. Then, an orthogonal linear transformation is constructed by simultaneously diagonalizing the CPMs of two tasks. Finally, each band's logarithmic power of the transformed signals is extracted for the support vector machine (SVM) classifier. The experiment results on multiple datasets showed that the CCSP algorithm is fully applicable to multi-channel EEG for reliable multi-cognitive-task identification.

Unsupervised double weight graphs based discriminant analysis for dimensionality reduction

ABSTRACTDimensionality reduction plays an important role in pattern recognition tasks. Locality preserving projection and neighbourhood preserving embedding are popular unsupervised feature extraction methods, which try to preserve a certain local structure in the low-dimensional subspace. However, only considering the local neighbour information will limit the methods to achieve higher recognition accuracy. In this paper, an unsupervised double weight graphs based discriminant analysis method (uDWG-DA) is proposed. First, uDWG-DA considers both similar and dissimilar relationships among samples by using double weight graphs. In order to explore the dissimilar information, a new partitioning strategy is proposed to divide the data set into different clusters, where samples of different clusters are dissimilar. Then, based on L2,1 norm, uDWG-DA finds the optimal projection to not only preserve the similar local structure but also increase the separability among different clusters of the data set. Experiments on four hyperspectral images validate the advantage and feasibility of the proposed method compared with other dimensionality reduction methods.

Food Safety Event Detection Based on Multi-Feature Fusion

Food safety event detection is a technique used to discover food safety events by monitoring online news. In general, a set of keywords are extracted as features to represent news, and then the news is clustered to generate events. The most popular method for news feature extraction is Term Frequency-Inverse Document Frequency (TF-IDF), however, it has some defects such as being prone to the “dimension disaster”, low computational efficiency, and a lack of semantic information. In addition, Latent Dirichlet Allocation (LDA) is also widely used in news representation. Despite its low dimension, it still suffers from some drawbacks such as the need to set a predefined number of clusters and has difficulty recognizing new events. In this paper, a method based on multi-feature fusion is proposed, which combines the TF-IDF features, the named entity features, and the headline features to represent the news. Based on the representations, the incremental clustering method is used to cluster the news documents and to detect food safety events. Compared with the traditional methods, the proposed method achieved higher Precision, Recall, and F1 scores. The proposed method can help regulatory authorities to make decisions and improve the reputation of the government, whilst reducing social anxiety and economic losses.

Local gradient of gradient pattern: a robust image descriptor for the classification of brain strokes from computed tomography images

This paper presents a new feature extraction method for the classification of brain computed tomography (CT) scan images into hemorrhagic strokes, ischemic strokes and normal CT images. The most popular feature extraction method is local binary pattern (LBP), which works by thresholding the neighboring pixel values with the center pixel value of the image. Unlike LBP, our proposed method is based on comparing neighbors of center pixel and the mean of whole image intensities in the first step, and computing double gradients of local neighborhoods of a center pixel of the original image in x and y directions in the second step. Further, values obtained from the first step are compared with double gradients of neighbors in order to generate codes for the center pixel. We have also calculated the codes for the first step. Thereafter, histograms of all the codes are generated and finally concatenated to form a single feature vector. We termed this descriptor as the local gradient of gradient pattern. We have performed nine different experiments where images have been classified using various classifiers. The efficacy of our feature descriptor for image classification is identified by comparing it with seven different feature extraction methods. Performances of these methods are tested using metrics such as precision, true positive rate, false positive rate, F-measure and accuracies of the classifier. Results obtained show that our method is superior to other previous descriptors.

Identifying Stable Patterns over Time for Emotion Recognition from EEG

In this paper, we investigate stable patterns of electroencephalogram (EEG) over time for emotion recognition using a machine learning approach. Up to now, various findings of activated patterns associated with different emotions have been reported. However, their stability over time has not been fully investigated yet. In this paper, we focus on identifying EEG stability in emotion recognition. We systematically evaluate the performance of various popular feature extraction, feature selection, feature smoothing and pattern classification methods with the DEAP dataset and a newly developed dataset called SEED for this study. Discriminative Graph regularized Extreme Learning Machine with differential entropy features achieves the best average accuracies of 69.67 and 91.07 percent on the DEAP and SEED datasets, respectively. The experimental results indicate that stable patterns exhibit consistency across sessions; the lateral temporal areas activate more for positive emotions than negative emotions in beta and gamma bands; the neural patterns of neutral emotions have higher alpha responses at parietal and occipital sites; and for negative emotions, the neural patterns have significant higher delta responses at parietal and occipital sites and higher gamma responses at prefrontal sites. The performance of our emotion recognition models shows that the neural patterns are relatively stable within and between sessions.

Detection of Gaps in Concrete⁻Metal Composite Structures Based on the Feature Extraction Method Using Piezoelectric Transducers.

A feature extraction methodology based on lamb waves is developed for the non-invasive detection and prediction of the gap in concrete–metal composite structures, such as concrete-filled steel tubes. A popular feature extraction method, partial least squares regression, is utilised to predict the gaps. The data is collected using the piezoelectric transducers attached to the external surface of the metal of the composite structure. A piezoelectric actuator generates a sine burst signal, which propagates along the metal and is received by a piezoelectric sensor. The partial least squares regression is performed on the raw sensor signal to extract features and to determine the relationship between the signal and the gap size, which is then used to predict the gaps. The applicability of the developed system is tested on two concrete-metal composite specimens. The first specimen consisted of an aluminium plate and the second specimen consisted of a steel plate. This technique is able to detect and predict gaps as low as 0.1 mm. The results demonstrate the applicability of this technique for the gap and debonding detection in concrete-filled steel tubes, which are critical in determining the degree of composite action between concrete and metal.

A Rotating Machinery Fault Diagnosis Method Based on Feature Learning of Thermal Images

The rotating machinery plays a vital role in industrial systems, in which unexpected mechanical faults during operation can lead to severe consequences. For fault prevention, many fault diagnostic methods based on vibration signals are available in the literature. However, the vibration signals are obtained by using different types of sensors, which can cause sensor installation issues and damage the rotating machinery. In addition, this kind of data acquisition through vibration signal induces a large amount of signal noise during machine operation, which will challenge the later fault diagnosis. A recent fault detection method based on infrared thermography (IRT) for rotating machinery avoids these issues. However, the corresponding literature is limited by the fact that the characteristics of the manual design cannot characterize the fault completely so that the diagnostic accuracy cannot exceed the diagnostic method based on the vibration signals. This paper introduces a popular image feature extraction method into the fault diagnosis of rotating machinery based on IRT for the first time. First, capturing the IRT images of the rotating machinery in different states, and then two popular feature extraction methods for IRT images, bag-of-visual-word, and convolutional neural network, are tested in turn. Finally, the extracted features are classified to implement the automatic fault diagnosis. The developed method is applied to analyze the experimental IRT images collected from bearings, and the results demonstrate that the developed method is more effective than the traditional methods based on vibration signals.

Iterative Kernel Principal Component for Large-Scale Data Set

Abstract Kernel principal component analysis (KPCA) is a popular nonlinear feature extraction method that uses eigendecomposition techniques to extract the principal components in the feature space. Most of the existing approaches are not feasible for analyzing large-scale data sets because of extensive storage needs and computation costs. To overcome these disadvantages, an efficient iterative method for computing kernel principal components is proposed. First, the power iteration is used to compute the first eigenvalue and the corresponding eigenvector. Then Schur-Weilandt deflation is repeatedly applied to obtain other higher order eigenvectors. No computation and storage of the kernel matrix is involved in this procedure. Instead, each row of the kernel matrix is calculated sequentially through the iterations. Thus, the kernel principal components can be computed without relying on the traditional eigendecomposition. The space complexity of the proposed method is O(m), and the time complexity is also greatly reduced. We illustrate the effectiveness of our approach through a series of real data experiments.