Proposed Feature Extraction Approach Research Articles

Moving object detection via feature extraction and classification

Abstract Foreground segmentation (FS) plays a fundamental and important role in computer vision, but it remains a challenging task in dynamic backgrounds. The supervised method has achieved good results, but the generalization ability needs to be improved. To address this challenge and improve the performance of FS in dynamic scenarios, a simple yet effective method has been proposed that leverages superpixel features and a one-dimensional convolution neural network (1D-CNN) named SPF-CNN. SPF-CNN involves several steps. First, the coined Iterated Robust CUR (IRCUR) is utilized to obtain candidate foregrounds for an image sequence. Simultaneously, the image sequence is segmented using simple linear iterative clustering. Next, the proposed feature extraction approach is applied to the candidate matrix region corresponding to the superpixel block. Finally, the 1D-CNN is trained using the obtained superpixel features. Experimental results demonstrate the effectiveness of SPF-CNN, which also exhibits strong generalization capabilities. The average F1-score reaches 0.83.

Moving object detection via feature extraction and classification

Abstract Foreground segmentation (FS) plays a fundamental and important role in computer vision, but it remains a challenging task in dynamic backgrounds. The supervised method has achieved good results, but the generalization ability needs to be improved. To address this challenge and improve the performance of FS in dynamic scenarios, a simple yet effective method has been proposed that leverages superpixel features and a one-dimensional convolution neural network (1D-CNN) named SPF-CNN. SPF-CNN involves several steps. First, the coined Iterated Robust CUR (IRCUR) is utilized to obtain candidate foregrounds for an image sequence. Simultaneously, the image sequence is segmented using simple linear iterative clustering. Next, the proposed feature extraction approach is applied to the candidate matrix region corresponding to the superpixel block. Finally, the 1D-CNN is trained using the obtained superpixel features. Experimental results demonstrate the effectiveness of SPF-CNN, which also exhibits strong generalization capabilities. The average F1-score reaches 0.83.

Computerized diagnosis of knee osteoarthritis from x‐ray images using combined texture features: Data from the osteoarthritis initiative

AbstractThe prevalence of knee osteoarthritis (KOA) cases has witnessed a significant increase on a global scale in recent years, emphasizing the need for automated diagnostic computer systems to aid in early‐stage osteoarthritis (OA) diagnosis. The accurate characterization of knee KOA stages through feature extraction poses significant research challenges due to the complexity of identifying relevant attributes. In this study, the development of a KOA diagnostic system is presented, leveraging a combination of Gabor, and Tamura parameters using the Canonical Correlation Analysis algorithm. Two feature selection algorithms, namely Principal Component Analysis and Relief, were employed for KOA classification. Furthermore, various classifiers, including K‐Nearest Neighbors, AdaBoost, Bagging, and Random Forest, were used to assess the proposed feature extraction approach. The diagnostic system was assessed using a dataset comprising 688 x‐ray images sourced from the OA initiative (OAI) dataset, consisting of 344 images from healthy subjects (Grade 0) and 344 images from pathological patients (Grade 2). To mitigate overfitting, a 10‐fold cross‐validation method was utilized. The experimental results indicate that the combination of Tamura and Gabor parameters with the Random Forest classifier achieved remarkable performance in KOA diagnosis, yielding an accuracy of 94.59%, and an area under the curve of 98.3%. Notably, the combined Gabor and Tamura models exhibited superior performance compared to individual models, as well as existing techniques reported in the literature.

A New Method for Heart Disease Detection: Long Short-Term Feature Extraction from Heart Sound Data.

Heart sounds have been extensively studied for heart disease diagnosis for several decades. Traditional machine learning algorithms applied in the literature have typically partitioned heart sounds into small windows and employed feature extraction methods to classify samples. However, as there is no optimal window length that can effectively represent the entire signal, windows may not provide a sufficient representation of the underlying data. To address this issue, this study proposes a novel approach that integrates window-based features with features extracted from the entire signal, thereby improving the overall accuracy of traditional machine learning algorithms. Specifically, feature extraction is carried out using two different time scales. Short-term features are computed from five-second fragments of heart sound instances, whereas long-term features are extracted from the entire signal. The long-term features are combined with the short-term features to create a feature pool known as long short-term features, which is then employed for classification. To evaluate the performance of the proposed method, various traditional machine learning algorithms with various models are applied to the PhysioNet/CinC Challenge 2016 dataset, which is a collection of diverse heart sound data. The experimental results demonstrate that the proposed feature extraction approach increases the accuracy of heart disease diagnosis by nearly 10%.

Fish Spoilage Classification Based on Color Distribution Analysis of Eye Images

Fish contains many nutrients beneficial to human health, which makes fish an essential component of a healthy diet. Omega-3 fatty acids, primarily found in fresh fish, can play a critical role in protecting heart and brain health. Freshness is one of the most important quality criteria of the fish to be selected for consumption. It is known that there may be pathogenic bacteria and toxins to human health in fish that are not stored in the right conditions and transferred by wrong logistics methods. One of the widely used approaches for evaluating the freshness of fish is sensory, which would be highly subjective and error-prone. Moreover, sensory analysis is widespread and one of the fastest approaches for evaluating large quantities of fish. At that point, a computer-aided diagnostic system can accelerate the evaluation of the degree of spoilage, reduce the human resources required for this task, and minimize the possibility of spoiled fish consumption. In this study, a fully automated freshness assessment mechanism based on the analysis of digital eye images of fish is proposed. Accordingly, the unsupervised clustering approach was used for feature extraction, and each image was divided into three regions according to their color distribution. The freshness was evaluated according to the intensity difference between these clusters. The results show that the proposed feature extraction approach is highly distinctive for the discrimination of spoilage and can be used to distinguish fresh fish from spoiled fish using machine learning methods without supervision.

Comparative Study of Mfcc and Mel Spectrogram for Raga Classification Using CNN

Objectives: To perform a comparative study of the results of feature extraction done using two different methods, Mfcc and Mel spectrogram, and determine which method is more effective for implementing the CNN algorithm. Methods: This study uses the CNN model to classify ragas according to Indian classical music. Feature extraction, which is a major operation in the Music Information Retrieval (MIR) process, is done using Mfcc and Mel spectrogram methods. The major ragas chosen as subjects for feature extraction are Yaman, Bhairav, Bhairavi, Multani, and Dhanashree. Findings: After comparison and examination of results achieved from both techniques, we could conclude that the CNN model using the Mel spectrogram method outperforms the CNN model using Mfcc. Novelty: The majority of the research, we discovered was on Carnatic music. In contrast to earlier research, this research takes a novel approach by conducting experiments on a variety of Hindustani classical ragas which are different from other studies. Researchers interested in music as well as application users would benefit from this study. Our proposed feature extraction approach will be useful for initializing the CNN algorithm, which will help aspiring musicians recognize ragas and classify songs based on these ragas. Keywords: Raga identification; Music Information Retrieval; Feature extraction; Mfcc; Melspectrograms; CNN

Vision-based gait analysis for real-time Parkinson disease identification and diagnosis system

ABSTRACT Computer-assisted Parkinson’s disease-specific gait pattern recognition has gained more attention in the past decade due to its extensive application. In this research study, vision-based gait feature extraction is obtained from the observed skeleton points to support the real-time Parkinson disease prediction and diagnosis in the smart healthcare environment. So, a novel kernel-based principal component analysis (KPCA) is introduced for establishing respective feature extraction and dimensionality reduction on the patient’s video data. In this research study, a vision-based Parkinson disease identification system (VPDIS) is developed with a feature-weighted minimum distance classifier model to support the clinical assessment of Parkinson’s disease. At the time of experimentation, a steady-state walking style of the patient was captured using the cameras fixed in the smart healthcare environment. Then, the accumulated walking frames from the remote patients were transformed into the required binary silhouettes for the sake of noise minimisation and compression purpose. The resulting experimentation shows that the proposed feature extraction approach has significant improvements on the recognition of target gait patterns from the video-based gait analysis of Parkinson’s and normal patients. Accordingly, the proposed VPDIS using feature-weighted minimum distance classifier model provides better prediction time and classification accuracy against the existing healthcare systems that is developed using support vector machine and ensemble learning classifier models.

New Feature Extraction Approaches Based on Spatial Points for Visual-Only Lip-Reading

The act of speaking takes place as a result of the joint use of both the senses of vision and hearing. The visual senses of the event of speech play an important role in lip-reading, especially when the sound is distorted or inaccessible. Visual-only-based lip-reading is a more difficult problem than audio-image-based lip-reading problems. In this study, three new spatial feature approaches to visual-only lip-reading are presented. To test the proposed feature extraction approaches, three datasets named AVLetters2 consisting of letters, AVDigits consisting of digits, and AVLetAVDig consisting of a combination of these two were used. First of all, the facial elements and lips were separated and the lip borders were marked with 20 points. Then, based on these spatial points, feature vectors were obtained with the feature approaches named Symmetric Euclidean Distance (SED), Central Euclidean Distance (CED), and Triple Points Angles (TPA). Extracted feature vectors were given to the CNN-LSTM network and 26 characters and 10 digits were tried to be estimated. As a result of the findings, the best success results for AVLetters2, AVDigits, and AVLetAVDig datasets were obtained by the SED+CNN+LSTM method as 53.2, 81.6, 59.8, respectively. When compared with the studies in the literature on the same data set, it was seen that very high and successful results were obtained.

EEG motor imagery classification using dynamic connectivity patterns and convolutional autoencoder

Recently, Brain computer Interface (BCI), plays an important role in recognizing brain activities and rehabilitation. Motor imagery (MI) classification based on Electroencephalography (EEG) signal analysis has received a lot of attention for the purpose of movement intent recognition. In this paper, we propose a novel feature extraction method which leads to promising MI classification performance. We split the individual EEG channels’ time series into temporal blocks and compute connectivity matrices for each block using adaptive sparse representation (ASR). The connectivity matrices indicate the correlation among different channels’ temporal blocks. They construct the dynamic connectivity patterns which are 3D tensors. Then, kernel PCA or non-linear convolutional autoencoders are applied to these tensors to learn discriminatory representations. This way, we incorporate the joint spatial-temporal structure of the EEG signals in the acquired features. We use a Gradient boosting model for classification. It is shown that the proposed feature extraction approach can lead to promising EEG MI classification performance compared to the state-of-the-art methods.

A Hybrid CFS Filter and RF-RFE Wrapper-Based Feature Extraction for Enhanced Agricultural Crop Yield Prediction Modeling

The innovation in science and technical knowledge has prompted an enormous amount of information for the agrarian sector. Machine learning has risen with massive processing techniques to perceive new contingencies in agricultural development. Machine learning is a novel onset for the investigation and determination of unpredictable agrarian issues. Machine learning models actualize the need for scaling the learning model’s performance. Feature selection can impact a machine learning model’s performance by defining a significant feature subset for increasing the performance and identifying the variability. This paper explains a novel hybrid feature extraction procedure, which is an aggregation of the correlation-based filter (CFS) and random forest recursive feature elimination (RFRFE) wrapper framework. The proposed feature extraction approach aims to identify an optimal subclass of features from a collection of climate, soil, and groundwater characteristics for constructing a crop-yield forecasting machine learning model with better performance and accuracy. The model’s precision and effectiveness are estimated (i) with all the features in the dataset, (ii) with essential features obtained using the learning algorithm’s inbuilt ‘feature_importances’ method, and (iii) with the significant features obtained through the proposed hybrid feature extraction technique. The validation of the hybrid CFS and RFRFE feature extraction approach in terms of evaluation metrics, predictive accuracies, and diagnostic plot performance analysis in comparison with random forest, decision tree, and gradient boosting machine learning algorithms are found to be profoundly satisfying.

Cross Domain Mean Approximation for Unsupervised Domain Adaptation

Unsupervised Domain Adaptation (UDA) aims to leverage the knowledge from the labeled source domain to help the task of target domain with the unlabeled data. It is a key step for UDA to minimize the cross-domain distribution divergence. In this paper, we firstly propose a novel discrepancy metric, referred to as Cross Domain Mean Approximation (CDMA) discrepancy, to evaluate the distribution differences between source and target domains, which calculate the sum of the squares of the distances from the source and target domains to the mean of the other domain. Secondly, Joint Distribution Adaptation based on Cross Domain Mean Approximation (JDA-CDMA) is developed on the basis of CDMA to extract shared feature and simultaneously reduce the marginal and conditional distribution discrepancy between domains during the label refinement process. Thirdly, we construct a classifier utilizing CDMA metric and neighbor information. Finally, the proposed feature extraction approach and classifier are combined to realize transfer learning. Results from extensive experiments on five visual benchmarks including object, face, and digit images, show the proposed methods outperform the state-of-the-art unsupervised domain adaptation.

Type-2 Fuzzy PCA Approach in Extracting Salient Features for Molecular Cancer Diagnostics and Prognostics.

Machine learning is becoming a powerful tool for cancer diagnosis and prognosis based on classification using high dimensional molecular data. However, extracting classification features from high-dimensional datasets remains a challenging problem. Principal component analysis (PCA) is a widely used method for dimensionality reduction. However, it is well-known that PCA and most PCA-based feature extraction methods are sensitive to noise, which may affect the accuracy of the subsequent classification. To address this problem, here we have proposed a robust fuzzy principal component analysis (PCA) with interval type-2 (IT-2) fuzzy membership functions for feature extraction. We have tested the performance of three widely used classifiers using the features extracted by proposed approaches and other feature extraction methods - PCA-based feature extraction methods (i.e. conventional PCA and fuzzy PCA), linear discriminant analysis (LDA), and support vector machine recursive feature elimination (SVM-RFE). The proposed feature extraction approaches showed better performance on cancer transcriptome and proteome datasets.

An Efficient Feature Extraction Technique Based on Local Coding PSSM and Multifeatures Fusion for Predicting Protein-Protein Interactions.

Background:Increasing evidence has indicated that protein-protein interactions (PPIs) play important roles in various aspects of the structural and functional organization of a cell. Thus, continuing to uncover potential PPIs is an important topic in the biomedical domain. Although various feature extraction methods with machine learning approaches have enhanced the prediction of PPIs. There remains room for improvement by developing novel and effective feature extraction methods and classifier approaches to identify PPIs.Method:In this study, we proposed a sequence-based feature extraction method called LCPSSMMF, which combined local coding position-specific scoring matrix (PSSM) with multifeatures fusion. First, we used a novel local coding method based on PSSM to build a new PSSM (CPSSM); the advantage of this method is that it incorporated global and local feature extraction, which can account for the interactions between residues in both continuous and discontinuous regions of amino acid sequences. Second, we adopted 2 different feature extraction methods (Local Average Group [LAG] and Bigram Probability [BP]) to capture multiple key feature information by employing the evolutionary information embedded in the CPSSM matrix. Finally, feature vectors were acquired by using multifeatures fusion method.Result:To evaluate the performance of the proposed feature extraction approach, we employed support vector machine (SVM) as a prediction classifier and applied this method to yeast and human PPI datasets. The prediction accuracies of LCPSSMMF were 93.43% and 90.41% on the yeast and human datasets, respectively. Moreover, we also compared the proposed method with the previous sequence-based approaches on the yeast datasets by using the same SVM classifier. The experimental results indicated that the performance of LCPSSMMF significantly exceeded that of several other state-of-the-art methods. It is proven that the LCPSSMMF approach can capture more local and global discriminatory information than almost all previous methods and can function remarkably well in identifying PPIs. To facilitate extensive research in future proteomics studies, we developed a LCPSSMMFSVM server, which is freely available for academic use at http://219.219.62.123:8888/LCPSSMMFSVM.

Data-driven damage diagnosis under environmental and operational variability by novel statistical pattern recognition methods

Feature extraction by time-series analysis and decision making through distance-based methods are powerful and efficient statistical pattern recognition techniques for data-driven structural health monitoring. The motivation of this article is to propose an innovative residual-based feature extraction approach based on AutoRegressive modeling and a novel statistical distance method named as Partition-based Kullback–Leibler Divergence for damage detection and localization by using randomly high-dimensional damage-sensitive features under environmental and operational variability. The key novel element of the proposed feature extraction approach is to establish a two-stage offline and online learning algorithms for extracting the residuals of AutoRegressive model as the main damage-sensitive features. This technique brings the great benefit of reducing the computational time and storage space for feature extraction in long-term monitoring conditions. The major contribution of Partition-based Kullback–Leibler Divergence method is to exploit a partitioning strategy for dividing random features into individual partitions and utilize numerical information of partitioning in distance calculation rather than directly applying random samples. Dealing with the major challenging issue of using the high-dimensional features in decision making and applicability to both correlated and uncorrelated random datasets are the main advantages of Partition-based Kullback–Leibler Divergence method. The accuracy and reliability of the proposed approaches are experimentally validated by two well-known benchmark structures. The stationarity and linearity of measured vibration responses for using in AutoRegressive modeling are evaluated by two hypothesis tests. Comparative studies are also conducted to demonstrate the superiority of the proposed methods over some exciting state-of-the-art techniques. Results show that the methods presented here succeed in detecting and locating damage and make time-saving and efficient tools for feature extraction and damage diagnosis.

Alzheimer's Disease Computer-Aided Diagnosis: Histogram-Based Analysis of Regional MRI Volumes for Feature Selection and Classification.

This paper proposes a novel fully automatic computer-aided diagnosis (CAD) system for the early detection of Alzheimer's disease (AD) based on supervised machine learning methods. The novelty of the approach, which is based on histogram analysis, is twofold: 1) a feature extraction process that aims to detect differences in brain regions of interest (ROIs) relevant for the recognition of subjects with AD and 2) an original greedy algorithm that predicts the severity of the effects of AD on these regions. This algorithm takes account of the progressive nature of AD that affects the brain structure with different levels of severity, i.e., the loss of gray matter in AD is found first in memory-related areas of the brain such as the hippocampus. Moreover, the proposed feature extraction process generates a reduced set of attributes which allows the use of general-purpose classification machine learning algorithms. In particular, the proposed feature extraction approach assesses the ROI image separability between classes in order to identify the ones with greater discriminant power. These regions will have the highest influence in the classification decision at the final stage. Several experiments were carried out on segmented magnetic resonance images from the Alzheimer's Disease Neuroimaging Initiative (ADNI) in order to show the benefits of the overall method. The proposed CAD system achieved competitive classification results in a highly efficient and straightforward way.

Feature Representation and Data Augmentation for Human Activity Classification Based on Wearable IMU Sensor Data Using a Deep LSTM Neural Network.

Wearable inertial measurement unit (IMU) sensors are powerful enablers for acquisition of motion data. Specifically, in human activity recognition (HAR), IMU sensor data collected from human motion are categorically combined to formulate datasets that can be used for learning human activities. However, successful learning of human activities from motion data involves the design and use of proper feature representations of IMU sensor data and suitable classifiers. Furthermore, the scarcity of labelled data is an impeding factor in the process of understanding the performance capabilities of data-driven learning models. To tackle these challenges, two primary contributions are in this article: first; by using raw IMU sensor data, a spectrogram-based feature extraction approach is proposed. Second, an ensemble of data augmentations in feature space is proposed to take care of the data scarcity problem. Performance tests were conducted on a deep long term short term memory (LSTM) neural network architecture to explore the influence of feature representations and the augmentations on activity recognition accuracy. The proposed feature extraction approach combined with the data augmentation ensemble produces state-of-the-art accuracy results in HAR. A performance evaluation of each augmentation approach is performed to show the influence on classification accuracy. Finally, in addition to using our own dataset, the proposed data augmentation technique is evaluated against the University of California, Irvine (UCI) public online HAR dataset and yields state-of-the-art accuracy results at various learning rates.

Email security level classification of imbalanced data using artificial neural network: The real case in a world-leading enterprise

Email is far more convenient than traditional mail in the delivery of messages. However, it is susceptible to information leakage in business. This problem can be alleviated by classifying emails into different security levels using text mining and machine learning technology. In this research, we developed a scheme in which a neural network is used to extract information from emails to enable its transformation into a multidimensional vector. Email text data is processed using bi-gram to train the document vector, which then undergoes under-sampling to deal with the problem of data imbalance. Finally, the security label of emails is classified using an artificial neural network. The proposed system was evaluated in an actual corporate setting. The results show that the proposed feature extraction approach is more effective than existing methods for the representations of email data in true positive rates and F1-scores.

A feature extraction method for predictive maintenance with time‐lagged correlation–based curve‐registration model

AbstractWith the prevalent development and use of predictive maintenance models for Internet‐of‐Things scenarios, the deep learning technology is gaining momentum. Feature extraction helps to increase efficiency in training the deep‐learning–based predictive maintenance model. However, there are common situations of time‐lagged correlations among industrial sensor data, resulting in reduction the effect of feature extraction. In this paper, we propose a feature extraction method for multisensors data with time‐lagged correlation. A curve‐registration method of correlation maximization algorithm is used to solve the problem of time‐lagged correlation for multi sensors. Then we apply a recurrent neural network, namely, long short‐term memory to develop a lightweight predictive maintenance model with the help of proposed feature extraction method. The effectiveness of the proposed feature extraction approach is demonstrated by examining real cases in a power plant. The experimental results indicate that our method can (1) effectively improve the accuracy of prediction and (2) improve the performance of the prediction model.

Rotation-invariant fingerprint matching using radon and DCT

A new set of promising rotation-invariant features based on radon and discrete cosine transform (DCT) is proposed for fingerprint matching. The radon and DCT of a tiny area in the region of core point of fingerprint image is computed. In the proposed method only 34% DCT coefficients are used for feature extraction. Competency of this approach is tested on standard databases, namely FVC2002 and FVC2004. This approach provides 70% genuine acceptance rate (GAR) at ~0% false acceptance rate (FAR) and 95% GAR at 10% FAR on rotated and non-rotated databases, respectively. Experimental results prove that the proposed feature extraction approach is rotation invariant.

An anomaly detection system based on variable N-gram features and one-class SVM

Context: Run-time detection of system anomalies at the host level remains a challenging task. Existing techniques suffer from high rates of false alarms, hindering large-scale deployment of anomaly detection techniques in commercial settings.Objective: To reduce the false alarm rate, we present a new anomaly detection system based on a novel feature extraction technique, which combines the frequency with the temporal information from system call traces, and on one-class support vector machine (OC-SVM) detector.Method: The proposed feature extraction approach starts by segmenting the system call traces into multiple n-grams of variable length and mapping them to fixed-size sparse feature vectors, which are then used to train OC-SVM detectors.Results: The results achieved on a real-world system call dataset show that our feature vectors with up to 6-grams outperform the term vector models (using the most common weighting schemes) proposed in related work. More importantly, our anomaly detection system using OC-SVM with a Gaussian kernel, trained on our feature vectors, achieves a higher-level of detection accuracy (with a lower false alarm rate) than that achieved by Markovian and n-gram based models as well as by the state-of-the-art anomaly detection techniques.Conclusion: The proposed feature extraction approach from traces of events provides new and general data representations that are suitable for training standard one-class machine learning algorithms, while preserving the temporal dependencies among these events.