Nonlinear Feature Extraction Research Articles

A Novel Indoor Positioning System Using Kernel Local Discriminant Analysis in Internet‐of‐Things

WLAN based localization is a key technique of location‐based services (LBS) indoors. However, the indoor environment is complex; received signal strength (RSS) is highly uncertain, multimodal, and nonlinear. The traditional location estimation methods fail to provide fair estimation accuracy under the said environment. We proposed a novel indoor positioning system that considers the nonlinear discriminative feature extraction of RSS using kernel local Fisher discriminant analysis (KLFDA). KLFDA extracts location features in a well‐preserved kernelized space. In the new kernel featured space, nonlinear RSS features are characterized effectively. Along with handling of nonlinearity, KLFDA also copes well with the multimodality in the RSS data. By performing KLFDA, the discriminating information contained in RSS is reorganized and maximally extracted. Prior to feature extraction, we performed outlier detection on RSS data to remove any anomalies present in the data. Experimental results show that the proposed approach obtains higher positioning accuracy by extracting maximal discriminate location features and discarding outlying information present in the RSS data.

Autoencoders and recommender systems: COFILS approach

We propose incorporating autoencoders into COFILS.Autoencoder COFILS presents competitive results against state-of-the-art methods.A new non-linear way to solve CF problem rather than straightforward approaches. Collaborative Filtering to Supervised Learning (COFILS) transforms a Collaborative Filtering (CF) problem into classical Supervised Learning (SL) problem. Applying COFILS reduces data sparsity and makes it possible to test a variety of SL algorithms rather than matrix decomposition methods. Its main steps are: extraction, mapping and prediction. Firstly, a Singular Value Decomposition (SVD) generates a set of latent variables from a ratings matrix. Next, on the mapping phase, a new data set is generated where each sample contains a set of latent variables from a user and each rated item; and a target that corresponds the user rating for that item. Finally, on the last phase, a SL algorithm is applied. One problem of COFILS is its dependency on SVD, that is not able to extract non-linear features from data and it is not robust to noisy data. To address this problem, we propose switching SVD to a Stacked Denoising Autoencoder (SDA) on the first phase of COFILS. With SDA, more useful and complex representations can be learned in a neural network with a local denoising criterion. We test our novel technique, namely Autoencoder COFILS (A-COFILS), on MovieLens, R3 Yahoo! Music and Movie Tweetings data sets and compare to COFILS, as a baseline, and state of the art CF techniques. Our results indicate that A- COFILS outperforms COFILS for all the data sets and with an improvement up to 5.9%. Also, A-COFILS achieves the best result for the MovieLens 100k data set and ranks on the top three algorithms for these data sets. Thus, we show that our technique represents an advance on COFILS methodology, improving its results and making it a suitable method for CF problem.

A Fixed-Point Online Kernel Principal Component Extraction Algorithm

Kernel principal component analysis (KPCA) is a powerful and widely applied nonlinear feature extraction technique. However, as originally proposed, KPCA may be cumbersome or infeasible in large-scale datasets, which motivated the development of low-complexity iterative extraction algorithms, mainly aiming image processing applications. Recently, some online KPCA extraction algorithms were proposed, but most of them suffer from low-convergence speed. This paper proposes a new algorithm based on fixed-point iterative equations for KPCA extraction, expanding kernel components using a compact dictionary, dynamically built from data, according to a user-defined accuracy parameter. The algorithm relies on simple equations, can track nonstationary environments, and requires reduced storage, enabling its use in real-time applications operating in low-cost embedded hardware. Results involving open-access image datasets show improved accuracy and convergence speed, as well as permitted effective improvements in practical image applications, as compared to state-of-art online KPCA techniques.

An improved Laplacian Eigenmaps method for machine nonlinear fault feature extraction

In the feature extraction of mechanical fault detection field, manifold learning is one of the effective nonlinear techniques. In this paper, aiming for the situations of noise sensitivity to manifold learning algorithms, an improved Laplacian Eigenmap (I-LapEig) algorithm is proposed and applied to the process of fault feature extraction. The new method takes advantage of local principal component analysis to eliminate the influence of noise points by reconstructing the neighborhood relation amongst the samples, and maintain the global intrinsic manifold structure, which enhances the performance of the feature extraction. To determine the parameters of I-LapEig algorithm, an adaptive neighborhood choose approach is presented. The K-nearest neighbor classifier is also adopted to implement feature classification and recognition. The experimental results on S-curve, rotor bed data, and compressor fault data show that the new method can effectively improve the performance of noise reduction in the feature extraction process when compared with the conventional local linear embedding and Laplacian Eigenmaps.

ALGORITHM FOR THE DETECTION OF CONGESTIVE HEART FAILURE INDEX

This study documents our efforts to provide computer support for the diagnosis of congestive heart failure (CHF). That computer support takes the form of an index value. A high index value indicates a low probability of CHF, and an index value below a threshold of 25.6 suggests a high probability of CHF. To create that index, we have designed a sophisticated algorithm chain which takes electrocardiogram signals as input. The signals are pre-processed before they are sent to a range of nonlinear feature extraction algorithms. The top 10 feature extraction methods were used to create the CHF index. By using objective feature extraction algorithms, we avoid the problem of inter- and intra-observer variability. We observed that the nonlinear feature extraction methods reflect the nature of the human heart very well. That observation is based on the fact that the nonlinear features achieved low [Formula: see text]-values and high feature ranking criterion scores.

NONLINEAR ANALYSIS OF CORONARY ARTERY DISEASE, MYOCARDIAL INFARCTION, AND NORMAL ECG SIGNALS

In this study, we analyze nonlinear feature extraction methods in terms of their ability to support the diagnosis of coronary artery disease (CAD) and myocardial infarction (MI). The nonlinear features were extracted from electrocardiogram (ECG) signals that were measured from CAD patients, MI patients as well as normal controls.We tested 34 recurrence quantification analysis (RQA) features, 14 bispectrum, and 136 cumulant features. The features were extracted from 10,546 normal, 41,545 CAD, and 40,182 MI heart beats. The feature quality was assessed with Student’s [Formula: see text]-test and the [Formula: see text]-value was used for feature ranking.We found that nonlinear features can effectively represent the physiological realities of the human heart.

Within‐component and between‐component multi‐kernel discriminating correlation analysis for colour face recognition

The key problem of colour face recognition technique is how to take full advantage of the colour information and extract effective discriminating features. To solve this problem, the authors propose a novel non-linear feature extraction approach for colour face recognition, named dual multi-kernel discriminating correlation analysis, which separately maps different colour components of face images into different non-linear kernel spaces, and then implements multi-kernel learning and discriminant analysis with the correlation metric not only within each colour component but also between diverse components. Then, to choose the optimum kernel space for each colour component and select the most suitable colour space for their approach, they design a kernel selection strategy and a colour space selection strategy, respectively. Experimental results in the face recognition grand challenge version 2 and labelled faces in the wilds databases validate the effectiveness of the proposed approach and two strategies.

A data-driven dynamics simulation framework for railway vehicles

ABSTRACTThe finite element (FE) method is essential for simulating vehicle dynamics with fine details, especially for train crash simulations. However, factors such as the complexity of meshes and the distortion involved in a large deformation would undermine its calculation efficiency. An alternative method, the multi-body (MB) dynamics simulation provides satisfying time efficiency but limited accuracy when highly nonlinear dynamic process is involved. To maintain the advantages of both methods, this paper proposes a data-driven simulation framework for dynamics simulation of railway vehicles. This framework uses machine learning techniques to extract nonlinear features from training data generated by FE simulations so that specific mesh structures can be formulated by a surrogate element (or surrogate elements) to replace the original mechanical elements, and the dynamics simulation can be implemented by co-simulation with the surrogate element(s) embedded into a MB model. This framework consists of a series of techniques including data collection, feature extraction, training data sampling, surrogate element building, and model evaluation and selection. To verify the feasibility of this framework, we present two case studies, a vertical dynamics simulation and a longitudinal dynamics simulation, based on co-simulation with MATLAB/Simulink and Simpack, and a further comparison with a popular data-driven model (the Kriging model) is provided. The simulation result shows that using the legendre polynomial regression model in building surrogate elements can largely cut down the simulation time without sacrifice in accuracy.

A new kernel method for hyperspectral image feature extraction

Hyperspectral image provides abundant spectral information for remote discrimination of subtle differences in ground covers. However, the increasing spectral dimensions, as well as the information redundancy, make the analysis and interpretation of hyperspectral images a challenge. Feature extraction is a very important step for hyperspectral image processing. Feature extraction methods aim at reducing the dimension of data, while preserving as much information as possible. Particularly, nonlinear feature extraction methods (e.g. kernel minimum noise fraction (KMNF) transformation) have been reported to benefit many applications of hyperspectral remote sensing, due to their good preservation of high-order structures of the original data. However, conventional KMNF or its extensions have some limitations on noise fraction estimation during the feature extraction, and this leads to poor performances for post-applications. This paper proposes a novel nonlinear feature extraction method for hyperspectral images. Instead of estimating noise fraction by the nearest neighborhood information (within a sliding window), the proposed method explores the use of image segmentation. The approach benefits both noise fraction estimation and information preservation, and enables a significant improvement for classification. Experimental results on two real hyperspectral images demonstrate the efficiency of the proposed method. Compared to conventional KMNF, the improvements of the method on two hyperspectral image classification are 8 and 11%. This nonlinear feature extraction method can be also applied to other disciplines where high-dimensional data analysis is required.

Epileptic electroencephalogram recognition based on discrete S transform and permutation entropy

Electroencephalogram(EEG) analysis has important reference value in the diagnosis of epilepsy. The automatic classification of epileptic EEG can be used to judge the patient's situation in time,which is of great significance in clinical application. In order to solve the problem that the recognition accuracy is not high by using the single feature of EEG signals and avoid the influence of wavelet basis function selection on recognition results,a method of automatic discrimination of epileptic EEG signals based on S transform and permutation entropy is proposed. Firstly, the original signals are decomposed by discrete S transform, and then we calculate the fluctuation index of coefficients of each rhythm and combine the permutation entropy of EEG signals into a feature vector and use Real AdaBoost classifier to discriminate the epileptic EEG signals in muti-period. In this study, we used the epilepsy database from University of Bonn. Three groups of EEG signals, including the data of normal people with their eyes open, the data collected inside of the epileptic foci from patients during their interictal period and the data during their ictal period, were used to test effectiveness. The results of this study showed that the fluctuation index of each rhythm could be used to characterize the normal, interictal and ictal epileptic EEG signals effectively, and the recognition accuracy of multiple features was much higher than that of single feature. The average recognition accuracy could reach 98.13%. Compared with time-frequency feature extraction method or nonlinear feature extraction method only,the recognition accuracy was increased by more than 1.2% and 8.1% respectively, which was superior to the methods recorded in many other literatures. Therefore, this method has a good application prospect in diagnosis of epilepsy.

Kernel PCA feature extraction and the SVM classification algorithm for multiple-status, through-wall, human being detection

Ultra-wideband (UWB) radar with strong anti-jamming performance and high-range resolution can be used to separate multiple human targets in a complex environment. In recent years, through-wall human being detection with UWB radar has become relatively sophisticated. In this paper, the method of kernel principal component analysis (KPCA) feature extraction and the support vector machine (SVM) classification algorithm are applied to identify and classify the multiple statuses of through-wall human being detection. This method makes full use of the KPCA of powerful, nonlinear feature extraction and SVMs, which can solve the problem of multiple-status detection and nonlinear pattern recognition. The experimental data that come from KPCA feature extraction are used as input to the SVM classification algorithm, some of which are used to train the model and the others to test the model. Experimental results showed that KPCA feature extraction and the SVM classification algorithm effectively distinguished four statuses of through-wall human being detection and achieved the desired results.

Hyperspectral anomaly detection based on stacked denoising autoencoders

Hyperspectral anomaly detection (AD) is an important technique of unsupervised target detection and has significance in real situations. Due to the high dimensionality of hyperspectral data, AD will be influenced by noise, nonlinear correlation of band, or other factors that lead to the decline of detection accuracy. To overcome this problem, a method of hyperspectral AD based on stacked denoising autoencoders (AE) (HADSDA) is proposed. Simultaneously, two different feature detection models, spectral feature (SF) and fused feature by clustering (FFC), are constructed to verify the effectiveness of the proposed algorithm. The SF detection model uses the SF of each pixel. The FFC detection model uses a similar set of pixels constructed by clustering and then fuses the set of pixels by the stacked denoising autoencoders algorithm (SDA). The SDA is an algorithm that can automatically learn nonlinear deep features of the image. Compared with other linear or nonlinear feature extraction methods, the detection result of the proposed algorithm is greatly improved. Experiment results show that the proposed algorithm is an excellent feature learning method and can achieve higher detection performance.

An Adaptive Supervised Nonlinear Feature Extraction for Hyperspectral Imagery Classification

In this paper, an improved version of locally linear Embedding is proposed. In the proposed method, spectral correlation angle is invited to describe the distance between data points, which is expected to fit the hyperspectral image (HSI). The neighborhood graph of the data points is constructed based on supervised method. Different from traditional supervised feature extraction methods, the weight factors, which are used to control the transform, are adaptively achieved. In this way, the input arguments of original algorithm are not increased. To justify the effectiveness of the proposed method, experiments are conducted on two HSIs. Results show that the proposed method can improve the separability of HSI especially in low dimensions.

Distinct Patterns of Tropical Pacific SST Anomaly and Their Impacts on North American Climate

A neural-network-based cluster technique, the so-called self-organizing map (SOM), was performed to extract distinct sea surface temperature (SST) anomaly patterns during boreal winter. The SOM technique has advantages in nonlinear feature extraction compared to the commonly used empirical orthogonal function analysis and is widely used in meteorology. The eight distinguishable SOM patterns so identified represent three La Niña–like patterns, two near-normal patterns, and three El Niño–like patterns. These patterns show the varied amplitude and location of the SST anomalies associated with El Niño and La Niña, such as the central Pacific (CP) and eastern Pacific (EP) El Niño. The impact of each distinctive SOM pattern on winter-mean surface temperature and precipitation changes over North America was examined. Based on composite maps with observational data, each SOM pattern corresponds to a distinguishable spatial structure of temperature and precipitation anomaly over North America, which seems to result from differing wave train patterns, extending from the tropics to mid–high latitudes induced by longitudinally shifted tropical heating. The corresponding teleconnection as represented by the National Center for Atmospheric Research Community Atmospheric Model, version 4 (CAM4), was compared with the observational results. It was found that the 16-member ensemble average of the CAM4 experiments with prescribed SST can reproduce the observed atmospheric circulation responses to the different SST SOM patterns, which suggests that the circulation differences are largely SST driven rather than due to internal atmospheric variability.

A Smartphone Indoor Localization Algorithm Based on WLAN Location Fingerprinting with Feature Extraction and Clustering.

With the development of communication technology, the demand for location-based services is growing rapidly. This paper presents an algorithm for indoor localization based on Received Signal Strength (RSS), which is collected from Access Points (APs). The proposed localization algorithm contains the offline information acquisition phase and online positioning phase. Firstly, the AP selection algorithm is reviewed and improved based on the stability of signals to remove useless AP; secondly, Kernel Principal Component Analysis (KPCA) is analyzed and used to remove the data redundancy and maintain useful characteristics for nonlinear feature extraction; thirdly, the Affinity Propagation Clustering (APC) algorithm utilizes RSS values to classify data samples and narrow the positioning range. In the online positioning phase, the classified data will be matched with the testing data to determine the position area, and the Maximum Likelihood (ML) estimate will be employed for precise positioning. Eventually, the proposed algorithm is implemented in a real-world environment for performance evaluation. Experimental results demonstrate that the proposed algorithm improves the accuracy and computational complexity.

Local kernel nonparametric discriminant analysis for adaptive extraction of complex structures

Abstract The linear discriminant analysis (LDA) is one of popular means for linear feature extraction. It usually performs well when the global data structure is consistent with the local data structure. Other frequently-used approaches of feature extraction usually require linear, independence, or large sample condition. However, in real world applications, these assumptions are not always satisfied or cannot be tested. In this paper, we introduce an adaptive method, local kernel nonparametric discriminant analysis (LKNDA), which integrates conventional discriminant analysis with nonparametric statistics. LKNDA is adept in identifying both complex nonlinear structures and the ad hoc rule. Six simulation cases demonstrate that LKNDA have both parametric and nonparametric algorithm advantages and higher classification accuracy. Quartic unilateral kernel function may provide better robustness of prediction than other functions. LKNDA gives an alternative solution for discriminant cases of complex nonlinear feature extraction or unknown feature extraction. At last, the application of LKNDA in the complex feature extraction of financial market activities is proposed.

Assessing the Scaling Subtraction Method for Impact Damage Detection in Composite Plates

The scaling subtraction method (SSM) is a non-destructive measurement approach used to extract nonlinear features from the elastic response of a structure. As such it can be used for damage detection purposes by identifying nonlinearities that may result from the presence of micro cracks or inclusions in granular and metallic materials. The effectiveness of such a technique to detect the presence of damage modes typical of laminated composite materials has not been yet assessed. With the purpose of filling this gap, in this paper the SSM is applied to inspect two laminated composite plates with different sizes, impact positions and sensor arrangement. Intact and damaged specimens are tested under harmonic excitations of different amplitude and frequency (the latter selected among the ultrasonic natural frequencies of the two plates). For each excitation case the recorded vibration signals are subtracted from the linearly rescaled reference signals and the SSM nonlinear indicators are calculated. The sensitivity of the method to the presence of damage is assessed in different sensor-receiver scenarios as well as for different excitation frequency and amplitude levels. Finite element numerical investigations are also performed to make comparisons with the experimental results.

Dual multi-kernel discriminant analysis for color face recognition

With the increasing use of color images in the fields of pattern recognition, computer vision and machine learning, color face recognition technique becomes important, whose key problem is how to make full use of the color information and extract effective discriminating features. In this paper, we propose a novel nonlinear feature extraction approach for color face recognition, named dual multi-kernel discriminant analysis (DMDA), where we design a kernel selection strategy to select the optimal kernel mapping function for each color component of face images, further design a color space selection strategy to choose the most suitable space, then separately map different color components of face images into different high-dimensional kernel spaces, and finally perform multi-kernel learning and discriminant analysis not only within each component but also between different components. Experimental results in the public face recognition grand challenge (FRGC) version 2 and labeled faces in the wilds (LFW) databases illustrate that our approach outperforms several representative color face recognition methods.

Selecting optimum base wavelet for extracting spectral alteration features associated with porphyry copper mineralization using hyperspectral images

Extracting a set of meaningful spectral features could enhance the classification performance. This is particularly important in hyperspectral images where the dataset are very large and time consuming to process. Wavelet transform as a powerful decomposition tool in both low and high frequency components could play an essential role in extracting spectral features of target minerals. Selecting the optimum base wavelet is an important step in wavelet transform. In this research, two criteria to select optimum base wavelet were implemented on three wavelet series including Daubechie (db), symlet (sym) and coiflet (coif). Energy criterion involves entropy factor and energy-to-Shannon entropy ratio while matching shape criterion operates according to correlation coefficients. High ranking base wavelets in both energy and shape criteria, coif1, db3 and db7, are recommended to be utilized in hyperspectral image classification. Neural Network technique was used for classification and trained by means of mineral spectral features related to typical porphyry copper deposits. Non-Linear wavelet feature extraction was employed to select the efficient features as input data. The study area covered by Hyperion data contains two well-known porphyry copper deposits, Darrehzar and Sarcheshmeh, located in the Iranian copper belt. Based on classification error matrix, it is concluded that db7 through 12 selected features exhibits the maximum consistency with the field measured data and can be recommended as an appropriate base wavelet for detecting porphyry copper deposits.

DNN Transfer Learning Based Non-Linear Feature Extraction for Acoustic Event Classification

Recent acoustic event classification research has focused on training suitable filters to represent acoustic events. However, due to limited availability of target event databases and linearity of conventional filters, there is still room for improving performance. By exploiting the non-linear modeling of deep neural networks (DNNs) and their ability to learn beyond pre-trained environments, this letter proposes a DNN-based feature extraction scheme for the classification of acoustic events. The effectiveness and robustness to noise of the proposed method are demonstrated using a database of indoor surveillance environments.