Integrated Feature Extraction Research Articles

Epilepsy Detection Method Based on the Time-gated Feature Network

Epilepsy is a nervous system disease, which is caused by abnormal discharge of brain neurons. The clinical manifestations are generalized seizures, clonus, loss of consciousness, and shock. An electroencephalogram (EEG) can accurately capture the changes in EEG activities. Therefore, EEG signals are used to detect seizures. In this paper, an epilepsy detection model based on a time-gated feature network (TFGN) is proposed. Firstly, the original EEG signal is preprocessed, and the preprocessed signal is sent into the TFGN detection model which integrates feature extraction, feature selection, and classification to obtain the detection results of epilepsy. Through the verification of data from different ages and channels, the detection accuracy of the TFGN detection model is higher than that of the traditional detection model, and the validity and comprehensiveness of the TFGN detection model are verified.

Visual similarity-based phishing detection using deep learning

Phishing is a common cybercrime that tempts a target to reveal personal and financial information by imitating a legitimate webpage. Classical techniques, such as list-based, and heuristic approaches have focused on identifying the uniform resource locator (URL). However, cybercriminals have uncovered several methods to evade URL-based techniques. Recently, researchers have started developing techniques that are based on visual similarity to detect phishing webpages. However, using visual similarity raises several computer vision-related issues such as webpage segmentation and feature extraction. Deep learning provides a suitable solution to these issues as it integrates feature extraction, feature selection, and classification in an end-to-end system. We have made the first attempt to detect phishing webpages based on visual similarity by modifying and retraining deep neural networks. A comprehensive evaluation on two publicly available databases has shown a marked improvement of around 7% classification accuracy.

Fake news detection using deep learning integrating feature extraction, natural language processing, and statistical descriptors

AbstractFake news potentially causes serious problems in society. Therefore, it is necessary to detect such news, which is, of course, associated with some challenges such as events, verification and datasets. Reference datasets related to this area face various problems, like the lack of sufficient information about news samples, no subject diversity, etc. The present paper proposes a model using feature extraction and machine learning algorithms for dealing with some of these problems. In the feature extraction phase, two new features (named coherence and cohesion), along with other key features, were extracted from news samples. In the detection phase, initially, the news samples of each dataset were sorted based on a specific order (easier samples in the beginning and harder ones towards the end) using a hybrid method consisting of statistical descriptors and a k‐nearest neighbor algorithm. Then, inspired by the human learning principles, the sorted news samples, were sent to the Long‐Short‐Term Memory and classical machine learning algorithms for the detection of fake news. The obtained results indicated the higher performance of the proposed model in fake news detection compared to benchmark models.

Task Encoding With Distribution Calibration for Few-Shot Learning

Few-shot learning is an extremely challenging task in computer vision that has attracted increased research attention in recent years. However, most recent methods do not fully use the task’s information, and few of the seen samples result in large intraclass differences among the same classes. In this paper, we propose a novel task encoding with distribution calibration (TEDC) model for few-shot learning, which uses the relationships among the feature distributions to reduce intraclass differences. In the TEDC model, an integrated feature extraction module (IFEM) is proposed, which extracts the multiangle visual features of an image and fuses them to obtain more representative features. To effectively utilize the task information, a novel task encoding module (TEM) is proposed, which obtains the task features by fusing all the seen samples’ information and uses them to adjust all the samples’ features for more generalizable task-specific representations. We also propose a distribution calibration module (DCM) to reduce the bias between the distribution of the support features and the query features in the same class. Extensive experiments show that our proposed TEDC model achieves an excellent performance and outperforms the state-of-the-art methods on three widely used few-shot classification benchmarks, specifically miniImageNet, tieredImageNet and CUB-200-2011.

Exploration of Pattern Recognition Methods for Motor Imagery EEG Signal with Convolutional Neural Network Approach

As an application of EEG, Motor Imagery based Brain-Computer Interface (MI BCI) plays a significant role in assisting patients with disability to communicate with their environment. MI BCI could now be realized through various methods such as machine learning. Many attempts using different machine learning approaches as MI BCI applications have been done with every one of them yielding various results. While some attempts managed to achieve agreeable results, some still failed. This failure may be caused by the separation of the feature extraction and classification steps, as this may lead to the loss of information which in turn causes lower classification accuracy. This problem can be solved by integrating feature extraction and classification by harnessing a classification algorithm that processed the input data as a whole until it produces the prediction, hence the use of convolutional neural network (CNN) approach which is known for its versatility in processing and classifying data all in one go. In this study, the CNN exploration involved a task to classify 5 different classes of fingers’ imaginary movement (thumb, index, middle, ring, and pinky) based on the processed raw signal provided. The CNN performance was observed for both non-augmented and augmented data with the data augmentation techniques used include sliding window, noise addition, and the combination of those two methods. From these experiments, the results show that the CNN model managed to achieve an averaged accuracy of 47%, meanwhile with the help of augmentation techniques of sliding window, noise addition, and the combined methods, the model achieved even higher averaged accuracy of 57,1%, 47,2%, and 57,5% respectively.

Airline ticket price-prediction model based on integrated feature extraction

Different feature extraction techniques are used to build AirFare-FS model, which is an integrated ticket price-prediction model, to solve the nonlinear regression problem of ticket price-prediction. Using three public air ticket datasets as an example, the AirFare-FS model identify main features affecting the air ticket price in each dataset and constructs a feature subset of each dataset using eleven feature extraction methods. Then, the AirFare-FS model selects the best feature subset of each dataset using a multi-objective optimization method. Finally, the optimal subset is used to find the best prediction method with the highest matching degree, and the dynamic adaptive model is constructed. The results show that the best feature subset of SixAirlines and EaseMyTrip datasets is subset 4 and the best matching prediction model is gradient descent, while the best subset of flight prices is subset 3 and the best matching prediction model is random forest. The visualization technology is used to show the effect of the characteristics of each optimal feature subset on the ticket price. The results indicate that the flight time dominantly affects the ticket price.

A Framework for Designing Unsupervised Pothole Detection by Integrating Feature Extraction Using Deep Recurrent Neural Network

A Framework for Designing Unsupervised Pothole Detection by Integrating Feature Extraction Using Deep Recurrent Neural Network

Interactive Oral English Chinese Machine Translation based on Feature Extraction Algorithm in accordance with the biotechnological advancement

The existing translation methods have the problem of imperfect prediction model, which leads to high KSMR value. This paper designs an interactive English Chinese oral machine translation method based on feature extraction algorithm using the management techniques of biotechnology field. The mathematical features of language are extracted, the trestle search method is used for pruning, the human-computer interaction interface is set up, the list of word candidate translation items is loaded, the interactive English Chinese oral machine translation prediction model is constructed, the linear interpolation is used for approximate solution, the feature extraction algorithm is used to design decoding strategy, and the discrimination of feature items for text topics is measured. Experimental results: compared with the other two translation methods, the mean value of KSMR of the designed translation method is 38.87%, 48.18% and 48.51%, which proves that the translation method integrating feature extraction algorithm has higher practical application value.

Intrusion Detection in Cyber Space Using Machine Learning Based Algorithm

Now a day, the fast growth of Internet access and the adoption of smart digital technology has resulted in new cybercrime strategies targeting regular people and businesses. The Web and social activities take precedence in most aspects of their lives, but also poses significant social risks. Static and dynamic analysis are inefficient in detecting unknown malware in standard threat detection approaches. Virus makers create new malware by modifying current malware using polymorphic and evasion tactics in order to fool. Furthermore, by utilizing selection of features techniques to identify more important features and minimizing amount of the data, these Machine Learning models' accuracy can be increased, resulting in fewer calculations. In the previous study traditional machine learning approaches were used to detect Malware. We employed Cuckoo sandbox, a malware detection and analysis system for detection and categorization, in this study we provide a Machine Learning based Intrusion analysis system to calculate exact and on spot Intrusion classification. We integrated feature extraction and component selection from the file, as well as selecting the much higher quality, resulting in exceptional accuracy and cheaper computing costs. For reliable identification and fine-grained categorization, we use a variety of machine learning algorithms. Our experimental results show that we achieved good, classified accuracy when compared to state-of-the-art approaches. We employed machine learning techniques such as K-Nearest Neighbor, Random Forest, Support Vector Machine, and Decision Tree. Using the Random Forest classifier on 108 features, we attained the greatest accuracy of 99.37 percent. We also discovered that Random Forest outscored all other classic machine learning techniques during the procedure. These findings can aid in the exact and accurate identification of Malware families.

An Unsupervised Feature Extraction Approach Based on Self-Expression.

Feature extraction algorithms lack good interpretability during the projection learning. To solve this problem, an unsupervised feature extraction algorithm, that is, block diagonal projection (BDP), based on self-expression is proposed. Specifically, if the original data are projected into a low-dimensional subspace by a feature extraction algorithm, although the data may be more compact, the new features obtained may not be as explanatory as the original sample features. Therefore, by imposing L2,1 norm constraint on the projection matrix, the projection matrix can be of row sparsity. On one hand, discriminative features can be selected to make the projection matrix to be more interpretable. On the other hand, irrelevant or redundant features can be suppressed. The proposed model integrates feature extraction and selection into one framework. In addition, since self-expression can well excavate the correlation between samples or sample features, the unsupervised feature extraction task can be better guided using this property between them. At the same time, the block diagonal representation regular term is introduced to directly pursue the block diagonal representation. Thus, the accuracy of pattern recognition tasks such as clustering and classification can be improved. Finally, the effectiveness of BDP in linear dimensionality reduction and classification is proved on various reference datasets. The experimental results show that this algorithm is superior to previous feature extraction counterparts.

A Systematic Deep Learning Model Selection for P300-Based Brain–Computer Interfaces

Predicting attention-modulated brain responses is a major area of investigation in brain&#x2013;computer interface (BCI) research that aims to translate neural activities into useful control and communication commands. Such studies involve collecting electroencephalographic (EEG) data from subjects to train classifiers for decoding users&#x2019; mental states. However, various sources of inter or intrasubject variabilities in brain signals render training classifiers in BCI systems challenging. From a machine learning perspective, this model training generally follows a common methodology: 1) apply some type of feature extraction, which can be time-consuming and may require domain knowledge and 2) train a classifier using extracted features. The advent of deep learning technologies has offered unprecedented opportunities to not only construct remarkably accurate classifiers but also to integrate the feature extraction stage into the classifier construction. Although integrating feature extraction, which is generally domain-dependent, into the classifier construction is a considerable advantage of deep learning models, the process of architecture selection for BCIs generally depends on domain knowledge. In this study, we examine the feasibility of conducting a systematic model selection combined with mainstream deep learning architectures to construct accurate classifiers for decoding P300 event-related potentials. In particular, we present the results of 232 convolutional neural networks (CNNs) (4 datasets <inline-formula> <tex-math notation="LaTeX">$\times58$ </tex-math></inline-formula> structures), 36 long short-term memory cells (LSTMs) (4 datasets <inline-formula> <tex-math notation="LaTeX">$\times9$ </tex-math></inline-formula> structures), and 320 hybrid CNN-LSTM models (4 datasets <inline-formula> <tex-math notation="LaTeX">$\times80$ </tex-math></inline-formula> structures) of varying complexity. Our empirical results show that in the classification of P300 waveforms, the constructed predictive models can outperform the current state-of-the-art deep learning architectures, which are partially or entirely inspired by domain knowledge. The source codes and constructed models are available at <uri>https://github.com/berdakh/P3Net</uri>.

Sports Target Tracking Based on Discriminant Correlation Filter and Convolutional Residual Network

During the sports tracking process, a moving target often encounters sophisticated scenarios such as fast motion and occlusion. During this period, erroneous tracking information will be generated and delivered to the next frame for updating; the information will seriously deteriorate the overall tracking model. To address the problem mentioned above, in this paper, we propose a convolution residual network model based on a discriminative correlation filter. The proposed tracking method uses discriminative correlation filters as basic convolutional layers in convolutional neural networks and then integrates feature extraction, response graph generation, and model updates into end-to-end convolutional neural networks for model training and prediction. Meanwhile, the introduction of residual learning responds to the model failure due to changes in the target appearance during the tracking process. Finally, multiple features are integrated such as HOG (histogram of oriented gradient), CN (color names), and histogram of local intensities for comprehensive feature representation, which further improve the tracking performance. We evaluate the performance of the proposed tracker on MultiSports datasets; the experimental results demonstrate that the proposed tracker performs favorably against most state-of-the-art discriminative correlation filter-based trackers, and the effectiveness of the feature extraction of the convolutional residual network is verified.

Gaming behavior and brain activation using functional near-infrared spectroscopy, Iowa gambling task, and machine learning techniques.

IntroductionThe current study investigates the utilization and performance of machine learning (ML) algorithms in the cognitive task of finding the correlation between numerical parameters of the human brain activation during gaming. We hypothesize that our integrated feature extraction platform is able to distinguish between different psychosomatic conditions in the gaming process as measured by the functional near‐infrared brain imaging technique.MethodsFor demonstration, the decision‐making process was constructed in the experiment environment that combined gaming simulator, such as the Iowa Gaming Task (IGT), with functional near‐infrared spectroscopy (fNIRS) as the neuroimaging technique. Features of fNIRS levels were extracted, averaged, and synchronized by time with the IGT dataset to predict the task score inside ML algorithms, such as multiple regression, classification and regression trees, support vector machine, artificial neural network, and random forest. For findings validation, the experiment data were resampled by training and testing sets. Further, a training dataset was used to train the ML algorithms, and prediction accuracy was estimated by repeated cross‐validation methods and compared by R squared and root mean square error (RMSE). The model with the best accuracy was used with the testing dataset and finalized the experiment.ResultsDuring the experiment, the highest correlation was identified in the fourth block between the oxy‐hemoglobin signal and IGT score in average value (0.24) and signal feature (0.57). Such relationship is due to block 4 characterization as “conceptual” period when participants task experience reaches the maximum, and rewards raise accordingly. Simultaneously, ML algorithms, constructed based on training data set, demonstrate acceptable performance, and RMSE as the primary performance metric dynamically increases from block 1 to block 5, from the state of uncertainty and unknown to the certainty and risky. In contrast, R squared decreases during the same transition. In most IGT blocks, the best fitted model was determined as support vector machine with radial bases function kernel, and predictions were made with the highest accuracy (lowest RMSE) than in training models.ConclusionObtained findings showed the applicability and capability of ML models as a powerful technique to evaluate the cognitive neuroimaging task result. Moreover, in terms of features it was identified that the hemodynamic response reacts to the acceleration decision‐making process and raises more significance than it was observed before.

On combining deep features and machine learning for automatic edge pedestrian detection task

Due to the restriction of cloud computing for video processing and distribution in networks, this paper designs an machine learning‐based automatic edge detection framework for video processing task in the Internet of Things (IoT) networks. Since the video processing algorithm is complex and consumes massive computing resources, it can reduce the network transmission to put the video processing model on the edge servers, uploading the processing results from the edge servers to the cloud computing center. Obviously, the processing result is much smaller than the original video, which can implement real‐time transmission. On the edge servers, the video is processed by utilizing a faster region convolutional neural network (Faster R‐CNN) which integrates feature extraction with classification. The experiments, performed on several pedestrian flow detection datasets, demonstrate the effectiveness of the designed automatic edge detection framework.

Matching Faces and Attributes Between the Artistic and the Real Domain: the PersonArt Approach

In this article, we present an approach for retrieving similar faces between the artistic and the real domain. The application we refer to is an interactive exhibition inside a museum, in which a visitor can take a photo of himself and search for a lookalike in the collection of paintings. The task requires not only to identify faces but also to extract discriminative features from artistic and photo-realistic images, tackling a significant domain shift. Our method integrates feature extraction networks which account for the aesthetic similarity of two faces and their correspondences in terms of semantic attributes. Also, it addresses the domain shift between realistic images and paintings by translating photo-realistic images into the artistic domain. Noticeably, by exploiting the same technique, our model does not need to rely on annotated data in the artistic domain. Experimental results are conducted on different paired datasets to show the effectiveness of the proposed solution in terms of identity and attribute preservation. The approach is also evaluated on unpaired settings and in combination with an interactive relevance feedback strategy. Finally, we show how the proposed algorithm has been implemented in a real showcase at the Gallerie Estensi museum in Italy, with the participation of more than 1,100 visitors in just three days.

Predicting the lifetime of Lithium–Ion batteries: Integrated feature extraction and modeling through sequential Unsupervised-Supervised Projections (USP)

Lithium-ion batteries are among the most used rechargeable batteries in the market, from portable electronics, and mobility solutions to process industry and aerospace. The ability to accurately predict battery lifetime and health status is of great interest for R&D activities, quality control & product release, predictive maintenance and recycling, among others. However, the complexity and multiplicity of degradation modes and usage/charging conditions raise considerable challenges to the construction of robust predictive models for battery lifecycle prediction. In this work, we propose an integrated methodology that is able to extract operational features and use them in the scope of a predictive model. Results demonstrate the superior feature extraction and accuracy of the proposed approach based only on the discharge information collected in the early usage periods (40–60 cycles), an aspect of practical interest. Data used regard commercial high-power LFP/graphite A123 cells, with nominal capacity of 1.1Ah and nominal voltage of 3.3 V.

Cotton Disease Detection Based on ConvNeXt and Attention Mechanisms

Cotton diseases cause low cotton production and fiber quality. Disease detection methods based on deep learning can integrate feature extraction and improve identification accuracy. We present an automatic cotton disease detection method to improve the identification accuracy of cotton disease. Cotton images are collected using a quadruped robot. ConvNeXt integrates the convolution neural network architecture with intrinsic superiority of transformer. The multiscale spatial pyramid attention (MSPA) module can help ConvNeXt concentrate on important regions of feature maps. ConvNeXt with the MSPA module shows the best recognition results of 97.2%, 99.7% and 100.0% on one competition dataset and two cotton datasets, respectively, with little increase in inference time. It indicates that the proposed model performs well in recognition accuracy with fast detection speed.

Online Automatic Diagnosis System of Cardiac Arrhythmias Based on MIT-BIH ECG Database.

Arrhythmias are a relatively common type of cardiovascular disease. Most cardiovascular diseases are often accompanied by arrhythmias. In clinical practice, an electrocardiogram (ECG) can be used as a primary diagnostic tool for cardiac activity and is commonly used to detect arrhythmias. Based on the hidden and sudden nature of the MIT-BIH ECG database signal and the small-signal amplitude, this paper constructs a hybrid model for the temporal correlation characteristics of the MIT-BIH ECG database data, to learn the deep-seated essential features of the target data, combine the characteristics of the information processing mechanism of the arrhythmia online automatic diagnosis system, and automatically extract the spatial features and temporal characteristics of the diagnostic data. First, a combination of median filter and bandstop filter is used to preprocess the data in the ECG database with individual differences in ECG waveforms, and there are problems of feature inaccuracy and useful feature omission which cannot effectively extract the features implied behind the massive ECG signals. Its diagnostic algorithm integrates feature extraction and classification into one, which avoids some bias in the feature extraction process and provides a new idea for the automatic diagnosis of cardiovascular diseases. To address the problem of feature importance variability in the temporal data of the MIT-BIH ECG database, a hybrid model is constructed by introducing algorithms in deep neural networks, which can enhance its diagnostic efficiency.

Integrating feature extraction approaches with hybrid emotional neural networks for water quality index modeling

The establishment of water quality prediction models is vital for aquatic ecosystems analysis. The traditional methods of water quality index (WQI) analysis are time-consuming and associated with a high degree of errors. These days, the application of artificial intelligence (AI) based models are trending for capturing nonlinear and complex processes. Therefore, the present study was conducted to predict the WQI in the Kinta River, Malaysia by employing the hybrid AI model i.e., GA-EANN (genetic algorithm-emotional artificial neural network). The extreme gradient boosting (XGB) and neuro-sensitivity analysis (NSA) approaches were utilized for feature extraction, and six different model combinations were derived to examine the relationship among the WQI with water quality (WQ) variables. The efficacy of the proposed hybrid GA-EANN model was evaluated against the backpropagation neural network (BPNN) and multilinear regression (MLR) models during calibration, and validation periods based on Nash–Sutcliffe efficiency (NSE), mean square error (MSE), root mean square error (RMSE), mean absolute percentage error (MAPE), and correlation coefficient (CC) indicators. According to the results of appraisal the hybrid GA-EANN model produced better outcomes (NSE = 0.9233/ 0.9018, MSE = 10.5195/ 9.7889 mg/L, RMSE = 3.2434/ 3.1287 mg/L, MAPE = 3.8032/ 3.0348 mg/L, and CC = 0.9609/ 0.9496) in calibration/ validation phases than BPNN and MLR models. In addition, the results indicate the better performance and suitability of the hybrid GA-EANN model with five input parameters in predicting the WQI for the study site.

LCHI: Low-Order Correlation and High-Order Interaction Integrated Model Oriented to Network Intrusion Detection

Network intrusion poses a severe threat to the Internet of Things (IoT). Thus, it is essential to study information security protection technology in IoT. Learning sophisticated feature interactions is critical in improving detection accuracy for network intrusion. Despite significant progress, existing methods seem to have a strong bias towards single low- or high-order feature interaction. Moreover, they always extract all possible low-order interactions indiscriminately, introducing too much noise. To address the above problems, we propose a low-order correlation and high-order interaction (LCHI) integrated feature extraction model. First, we selectively extract the beneficial low-order correlation between the same-type features by the multivariate correlation analysis (MCA) model and attention mechanism. Second, we extract the complicated high-order feature interaction by the deep neural network (DNN) model. Finally, we emphasize both the low- and high-order feature interactions and incorporate them. Our LCHI model seamlessly combines the linearity of MCA in modeling lower-order feature correlation and the nonlinearity of DNN in modeling higher-order feature interaction. Conceptually, our LCHI is more expressive than the previous models. We carry on a series of experiments on the public wireless and wired network intrusion detection datasets. The experimental results show that LCHI improves 1.06%, 2.46%, 3.74%, 0.25%, 1.17%, and 0.64% on the AWID, NSL-KDD, UNSW-NB15, CICIDS 2017, CICIDS 2018, and DAPT 2020 datasets, respectively.