Singular Vector Decomposition Research Articles

Face, Fingerprint, and Signature based Multimodal Biometric System using Score Level and Decision Level Fusion Approaches

Principal Component Analysis (PCA) is the best face recognition method. This research suggests PCA for fingerprint and signature recognition. Simple image processing transforms like DCT, 2D-DCT, DWT, SWT, 2D-SWT, SVD (Singular Vector Decomposition), Entropy, and Rank can be used for feature extraction. These transforms and measures are utilized with PCA as a feature extraction module to construct uni-modal and multimodal biometric systems using face, fingerprint, and signature modalities. Most PCA biometrics systems compare stored template to claimed identification using Euclidean distance. This paper proposes matching modules using similarity and dissimilarity measures viz. Absolute Pearson's Correlation Coefficient (APCC), Absolute Uncentered Pearson's Correlation Coefficient (AUPCC), Bray Curtis Distance (BC), Canberra distance (CB), Chebyshev Distance (CBS), Chessboard Distance (CSB), City block or Manhattan distance (CTB), Cross Correlation (CC), Dot product (DP), Euclidean distance (EUC), Extended Jaccard Distance (EJ), Hamming Distance (HM), Harmonically Summed Euclidean distance (HSEUC), Kendall Correlation Coefficient (KCC), Mahalanobis Distance (MH), Minimum Coordinate Difference (MCD), Minkowiski distance (MNK), Multivariate Kurtosis Coefficient (MVK), Multivariate Skew (MVS), Normalized City Block or Manhattan distance (NCTB), Normalized Cross-correlation (NCC), Normalized Euclidean distance (NEUC), Pearson’s Cosine Distance (PCOS), Pearson's Correlation Coefficient (PCC), Pearson's Absolute Value Dissimilarity (PAVD), Pearson's Linear Dissimilarity (PLDISS), Spearman Correlation Coefficient (SCC), Standardized Euclidean Distance (SEUC), Uncentered Pearson's Correlation Coefficient (UPCC), Wave-Hedges Distance (WVH). This study again discusses score level fusion of face, fingerprint, and signature using sum and max rules, z-score normalization, and decision level fusion using AND rule.

Breast Cancer Prediction Using Fine Needle Aspiration Features and Upsampling with Supervised Machine Learning.

Breast cancer is one of the most common invasive cancers in women and it continues to be a worldwide medical problem since the number of cases has significantly increased over the past decade. Breast cancer is the second leading cause of death from cancer in women. The early detection of breast cancer can save human life but the traditional approach for detecting breast cancer disease needs various laboratory tests involving medical experts. To reduce human error and speed up breast cancer detection, an automatic system is required that would perform the diagnosis accurately and timely. Despite the research efforts for automated systems for cancer detection, a wide gap exists between the desired and provided accuracy of current approaches. To overcome this issue, this research proposes an approach for breast cancer prediction by selecting the best fine needle aspiration features. To enhance the prediction accuracy, several feature selection techniques are applied to analyze their efficacy, such as principal component analysis, singular vector decomposition, and chi-square (Chi2). Extensive experiments are performed with different features and different set sizes of features to investigate the optimal feature set. Additionally, the influence of imbalanced and balanced data using the SMOTE approach is investigated. Six classifiers including random forest, support vector machine, gradient boosting machine, logistic regression, multilayer perceptron, and K-nearest neighbors (KNN) are tuned to achieve increased classification accuracy. Results indicate that KNN outperforms all other classifiers on the used dataset with 20 features using SVD and with the 15 most important features using a PCA with a 100% accuracy score.

Prediction of beta -Thalassemia carriers using complete blood count features

beta -Thalassemia is one of the dangerous causes of the high mortality rate in the Mediterranean countries. Substantial resources are required to save a beta -Thalassemia carriers’ life and early detection of thalassemia patients can help appropriate treatment to increase the carrier’s life expectancy. Being a genetic disease, it can not be prevented however the analysis of several indicators in parents’ blood can be used to detect disorders causing Thalassemia. Laboratory tests for Thalassemia are time-consuming and expensive like high-performance liquid chromatography, Complete Blood Count (CBC) with peripheral smear, genetic test, etc. Red blood indices from CBC can be used with machine learning models for the same task. Despite the available approaches for Thalassemia carriers from CBC data, gaps exist between the desired and achieved accuracy. Moreover, the data imbalance problem is studied well which makes the models less generalizable. This study proposes a highly accurate approach for beta -Thalassemia detection using red blood indices from CBC augmented by supervised machine learning. In view of the fact that all the features do not carry predictive information regarding the target variable, this study employs a unified framework of two features selection techniques including Principal Component Analysis (PCA) and Singular Vector Decomposition (SVD). The data imbalance between beta -Thalassemia carrier and non-carriers is handled by Synthetic Minority Oversampling Technique (SMOTE) and Adaptive Synthetic (ADASYN). Extensive experiments are performed using many state-of-the-art machine learning models and deep learning models. Experimental results indicate the superiority of the proposed approach over existing approaches with an accuracy score of 0.96.

HRP UWB 통신의 채널 특성 변화를 이용한 실내 사람 수 파악

HRP UWB technology has recently attracted attention owing to its advantages of enabling precise distance measurements as well as communication. In addition to communication and distance measurement, passive radar applications that can recognize the surrounding situation are also being considered. Among these radar applications, the ability to detect the number of people indoors is of increasing interest when social distancing is important. In this study, we implemented an HRP UWB passive radar application system that detects the number of people in an indoor environment using the change in channel characteristics between HRP UWB communication devices. Using the implemented system, changes in the channel impulse response (CIR) characteristics of channels 5 and 9 were observed, and the number of people was identified using the singular vector decomposition (SVD) algorithm. From actual measurements, it was possible to confirm that the number of people within five people in an indoor environment could be accurately identified using a pair of HRP UWB wireless communication devices in an indoor environment.

Recent advances in phase-array based wideband multimodal dispersion curves extraction and plate-waveguide parameter inversion

Multimodal ultrasonic guided waves with rich dispersion information have been widely applied for waveguide evaluation. However, there are still some challenges to extract multimode dispersion curves using the traditional pitch-catch method, including lack of spatial information and poor signal-to-noise ratio. Phase array has been recently developed for measuring ultrasonic guided waves, which brings many advantages for wideband multimodal dispersion curves extraction and parameter inversion. In the last years, basing on a multi-emitter and multi-receiver measurement, we proposed some array dispersive signal processing strategies for analyzing guided waves in long cortical bone, including the sparse singular vector decomposition (sparse-SVD) method for enhancing wavenumber estimation resolution and the low-amplitude mode extraction, Radon transform and dispersive Radon transform (DRT) with the capability to project temporal array dispersive signals on the space of parameters of interest for solving the inverse problem, and our recent work of deep neural networks for solving the intractable multiparameter inverse problem from the array signals to the waveguide elasticity. In the talk, we present (1) recent advances in array signal processing for extracting wideband multimodal dispersion curves; (2) some new perspectives to retrieve the waveguide parameters; (3) some pilot clinical results of long cortical bone evaluation using ultrasonic guided waves.

Harmonizing solar induced fluorescence across spatial scales, instruments, and extraction methods using proximal and airborne remote sensing: A multi-scale study in a soybean field

Solar-induced chlorophyll fluorescence (SIF) has been widely used to track vegetation photosynthesis at different scales ranging from in-situ measurements to satellite products. Airborne platforms sample SIF data at a spatial scale intermediate between in-situ and satellite, matching that of ground measurement (e.g. flux tower footprints and other field sampling), enabling us to explore causes of SIF variation and validate satellite-based SIF products. However, harmonizing SIF across sensors and platforms (correcting for systematic errors to yield a consistent, comparable SIF product) is challenging because SIF can be retrieved in different absorption windows, with different instruments and methods complicating the comparison between different observational levels (i.e., ground, airborne, satellites) and between sites equipped with different instruments with varying optical properties (spectral resolution and sampling intervals, spatial resolution). Additionally, the spatial and temporal variability of atmospheric properties can influence the retrieval of the weak SIF signal. Because of these complications, direct comparisons of airborne and ground SIF across scales are rarely attempted. In this study, we combined airborne SIF data with simultaneous ‘ground truth’ data collected by stationary and mobile platforms in a soybean field in Nebraska, USA. In this effort, we tested several SIF extraction methods, including Fraunhofer Line Discrimination (FLD), improved Fraunhofer Line Discrimination (iFLD), Spectral Fitting Method (SFM), SpecFit, and a Singular Vector Decomposition (SVD) method. The SpecFit method was sensitive to the 715–740 nm water bands and removing the water bands in the fitting process yielded better agreement between the airborne and ground SIF spectra. Accurate estimation of the ground level downwelling irradiance obtained by ground measurements over a calibration target improved agreement between airborne and ground SIF retrievals at the O2A band, and allowed us to derive a SIF dataset with improved agreement across platforms and sampling scales. This experimental approach provided a method for generating comparable SIF signals across instruments, methods and platforms, which is critical to understanding the SIF-GPP relationship at different scales and to cross-validate the diversity of platforms used for satellite products calibration and validation.

Randomized Spatial Downsampling-Based Cauchy-RPCA Clutter Filtering for High-Resolution Ultrafast Ultrasound Microvasculature Imaging and Functional Imaging.

Effective tissue clutter filtering and noise removing are essential for ultrafast Doppler imaging. Singular vector decomposition (SVD)-based spatiotemporal method has been applied as a classical method to remove the clutter and strong motion artifacts. However, performance of the SVD-based methods often depends on a proper eigenvector thresholding, i.e., the separation of signal subspaces of small-value blood flow, large-value static tissue, and noise. In the study, a Cauchy-norm-based robust principal component analysis (Cauchy-RPCA) method is developed via Cauchy-norm-based sparsity penalization, which enhances the blood flow extraction of small-vessels. A randomized spatial downsampling strategy and alternating direction method of multipliers (ADMM) are further involved to accelerate the computation. A face-to-face comparison is carried out among the classical SVD, traditional RPCA, blind deconvolution-based RPCA (BD-RPCA), and the proposed Cauchy-RPCA methods. Ultrafast ultrasound imaging dataset recorded from rat brain is used to investigate the performance of the proposed Cauchy-RPCA method in terms of clutter filtering, power Doppler, color Doppler, and functional ultrasound (fUS) imaging. The computational efficiency is finally discussed.

CLPA: Clean-Label Poisoning Availability Attacks Using Generative Adversarial Nets

Poisoning attacks are emerging threats to deep neural networks where the adversaries attempt to compromise the models by injecting malicious data points in the clean training data. Poisoning attacks target either the availability or integrity of a model. The availability attack aims to degrade the overall accuracy while the integrity attack causes misclassification only for specific instances without affecting the accuracy of clean data. Although clean-label integrity attacks are proven to be effective in recent studies, the feasibility of clean-label availability attacks remains unclear. This paper, for the first time, proposes a clean-label approach, CLPA, for the poisoning availability attack. We reveal that due to the intrinsic imperfection of classifiers, naturally misclassified inputs can be considered as a special type of poisoned data, which we refer to as "natural poisoned data''. We then propose a two-phase generative adversarial net (GAN) based poisoned data generation framework along with a triplet loss function for synthesizing clean-label poisoned samples that locate in a similar distribution as natural poisoned data. The generated poisoned data are plausible to human perception and can also bypass the singular vector decomposition (SVD) based defense. We demonstrate the effectiveness of our approach on CIFAR-10 and ImageNet dataset over a variety type of models. Codes are available at: https://github.com/bxz9200/CLPA.

Spectral Graph Wavelet Based Image Steganography Using SVD and Arnold Transform

Steganography is the process of concealing sensitive information within cover medium. This study offers an efficient and safe innovative image steganography approach based on graph signal processing (GSP). To scramble the secret image, Arnold cat map transform is used, then Spectral graph wavelet is used to change the cover and scrambled secret image, followed by singular vector decomposition (SVD) of the modified cover image. To create the stego image, an alpha blending process is used. To produce the stego image, GSP-based synthesis is used. By maintaining the inter-pixel correlation, GSP improves the visual quality of the produced stego image. The effects of image processing attacks on the suggested approach are examined. The investigational results and assessment indicate that the proposed steganography scheme is more efficient and robust in terms of quality measures. The quality of stego image is evaluated in respect of PSNR, NCC, SC and AD performance metrics.

Computation analysis of regularization methods and parameter selection for acoustics radiation modes source reconstruction of vibrating plates

Source localization and power estimation is a topic of great interest in acoustics and vibration. Acoustic source radiation modes reconstruction is a method of particular interest. Solutions leads to determinate sound/vibration power and surface velocity distribution from sparse acoustics samples. It has been shown that the quality of the results depends on Tikhonov regularization parameter. This inverse method is based on the radiation resistance matrix and we show that a single instruction multiple threads computing approach for graphics processing unit device exhibit better speed performance than common approaches to achieve the solution. We compare four regularization and three estimating methods for regularization parameters. We use a similarity measure to the simulated cases in three frequencies. Tikhonov regularization exhibits best reconstruction results. However, truncated singular vector decomposition also shows good performance with the advantage of not using a regularization parameter. Graphics processing unit implementation reduce reconstruction's computing time at least in a half.

Retrieving Sun-Induced Chlorophyll Fluorescence from Hyperspectral Data with TanSat Satellite.

A series of algorithms for satellite retrievals of sun-induced chlorophyll fluorescence (SIF) have been developed and applied to different sensors. However, research on SIF retrieval using hyperspectral data is performed in narrow spectral windows, assuming that SIF remains constant. In this paper, based on the singular vector decomposition (SVD) technique, we present an approach for retrieving SIF, which can be applied to remotely sensed data with ultra-high spectral resolution and in a broad spectral window without assuming that the SIF remains constant. The idea is to combine the first singular vector, the pivotal information of the non-fluorescence spectrum, with the low-frequency contribution of the atmosphere, plus a linear combination of the remaining singular vectors to express the non-fluorescence spectrum. Subject to instrument settings, the retrieval was performed within a spectral window of approximately 7 nm that contained only Fraunhofer lines. In our retrieval, hyperspectral data of the O2-A band from the first Chinese carbon dioxide observation satellite (TanSat) was used. The Bayesian Information Criterion (BIC) was introduced to self-adaptively determine the number of free parameters and reduce retrieval noise. SIF retrievals were compared with TanSat SIF and OCO-2 SIF. The results showed good consistency and rationality. A sensitivity analysis was also conducted to verify the performance of this approach. To summarize, the approach would provide more possibilities for retrieving SIF from hyperspectral data.

The Ambiguity Issue in Solving Inverse Problems of Small-Angle Scattering: A Consistent Approach Using an Insulin Receptor-Related Receptor as an Example. Methods for Interpreting SAXS Data

The construction of three-dimensional models of protein macromolecules is a serious challenge due to the possible ambiguity of solving the inverse problem of reconstructing a three-dimensional structure from a one-dimensional small-angle scattering profile. The target function of this task can have several local minima, which leads to the dependence of the solution on the initial values of the model parameters and on the method of finding the global minimum. The problem of creating structural models is also complicated by averaging the scattering pattern over all orientations of particles in space and by the size and shape distribution of scattering objects in the case of polydispersity and/or polymorphism. In this study, the issue of ambiguity in solving inverse problems and restoring the three-dimensional structure of a protein is considered using the structure of the ectodomain of an insulin receptor-related receptor (ectoIRR) in solution as an example. The paper presents a consistent approach to solving this problem, starting from the determination of general structural parameters and ab initio reconstruction of shape to modeling by rigid bodies (using molecular tectonics), hybrid methods, and analysis of scattering profiles by singular vector decomposition.

Nonlinear process monitoring using a mixture of probabilistic PCA with clusterings

Motivated by mixture of probabilistic principal component analysis (PCA), which is time-consuming due to expectation maximization, this paper investigates a novel mixture of probabilistic PCA with clusterings for process monitoring. The significant features are extracted by singular vector decomposition (SVD) or kernel PCA, and k-means is subsequently utilized as a clustering algorithm. Then, parameters of local PCA models are determined under each clustering model. Compared with PCA clustering, SVD based clustering only utilizes the nature basis for the components of the data instead of principal components of the data. Three clustering approaches are adopted and the effectiveness of the proposed approach is demonstrated by a practical coal pulverizing system.

Comparison of accuracy between LC model and 4-PFM when COVID-19 impacts mortality structure

This paper studies if the accuracies of mortality models (LC model vs. 4-parametric model) are aggravated if a mortality structure changes due to the impact of COVID-19. LC model (LCM) uses dimension reduction for fitting to the log mortality matrix so that the performance of the dimension reduction method may not be good when the matrix structure changes. On the other hand, 4-parametric factor model (4-PFM) is designed to use factors for fitting to log mortality data by age groups so that it would be less affected by the change of the mortality structure. In fact, the forecast accuracies of LCM are better than those of 4-PFM when life-tables are used whereas those of 4-PFM are better when the mortality structure changes. Thus this result shows that 4-PFM is more reliable in performance to the structural changes of the mortality. To support the accuracy changes of LCM the functional aspect is explained by computing eigenvalues produced by singular vector decomposition

Color edge detection by learning classification network with anisotropic directional derivative matrices

In this paper, a neural network-based color edge detector is constructed by learning a classifier using anisotropic directional derivative (ANDD) matrices of a color image as input. The training stage on a color edge dataset with ground truth (GT) edges includes calculation of ANDD matrices, generation of feature matrices, and training a classifier. For each training image, a set of ANDD matrices are calculated from the ANDDs with different parameter setups for training and from which a set of the color edge strength maps (CESMs) are extracted by the singular vector decomposition. The CESMs and the GTs on edges of the image are combined into a feature matrix for training. Using the feature matrices of all the training images as input, a classification neural network is trained and it outputs the probability of a pixel to be an edge pixel. In the detection stage, for a color image, its ANDD matrices, CESMs, and the color edge direction maps (CEDMs) are first computed and then the CESMs are input into the classification neural network to obtain the edge probability map (EPM) of the image. Finally, the non-maximum suppression and hysteresis thresholding are applied to the EPM and CEDMs to generate the binary edge map. The proposed detector attains better performance than the existing gradient-based detectors and is competitive with learning-based detectors on three commonly-used color image datasets for edge and contour detection.

Modified High-Order SVD for Spatiotemporal Modeling of Distributed Parameter Systems

Modeling high-spatial dimensional (high-D) distributed parameter systems (DPSs) is very difficult because of the spatially distributed characteristic and complex spatiotemporal coupling. In this article, a new framework based on high-order singular vector decomposition (HOSVD) is proposed to model a high-D DPS. A modified HOSVD is designed for separation of time and multiple spatial variables. It can well preserve the spatially distributed nature of the high-D DPS. It also considers the interaction across different spatial modes, which is an important part of spatiotemporal coupling in the high-D DPS. Experiments of a two-spatial dimensional thermal curing process are used to verify the effectiveness of the proposed method. An average prediction error near 1% of the curing temperature can be achieved.

Optimally growing initial errors of El Ni\xf1o events in the CESM

The optimally growing initial errors (OGEs) of El Niño events are found in the Community Earth System Model (CESM) by the conditional nonlinear optimal perturbation (CNOP) method. Based on the characteristics of low-dimensional attractors for ENSO (El Niño Southern Oscillation) systems, we apply singular vector decomposition (SVD) to reduce the dimensions of optimization problems and calculate the CNOP in a truncated phase space by the differential evolution (DE) algorithm. In the CESM, we obtain three types of OGEs of El Niño events with different intensities and diversities and call them type-1, type-2 and type-3 initial errors. Among them, the type-1 initial error is characterized by negative SSTA errors in the equatorial Pacific accompanied by a negative west–east slope of subsurface temperature from the subsurface to the surface in the equatorial central-eastern Pacific. The type-2 initial error is similar to the type-1 initial error but with the opposite sign. The type-3 initial error behaves as a basin-wide dipolar pattern of tropical sea temperature errors from the sea surface to the subsurface, with positive errors in the upper layers of the equatorial eastern Pacific and negative errors in the lower layers of the equatorial western Pacific. For the type-1 (type-2) initial error, the negative (positive) temperature errors in the eastern equatorial Pacific develop locally into a mature La Niña (El Niño)-like mode. For the type-3 initial error, the negative errors in the lower layers of the western equatorial Pacific propagate eastward with Kelvin waves and are intensified in the eastern equatorial Pacific. Although the type-1 and type-3 initial errors have different spatial patterns and dynamic growing mechanisms, both cause El Niño events to be underpredicted as neutral states or La Niña events. However, the type-2 initial error makes a moderate El Niño event to be predicted as an extremely strong event.

Ensemble Methods for Neural Network‐Based Weather Forecasts

AbstractEnsemble weather forecasts enable a measure of uncertainty to be attached to each forecast, by computing the ensemble's spread. However, generating an ensemble with a good spread‐error relationship is far from trivial, and a wide range of approaches to achieve this have been explored—chiefly in the context of numerical weather prediction models. Here, we aim to transform a deterministic neural network weather forecasting system into an ensemble forecasting system. We test four methods to generate the ensemble: random initial perturbations, retraining of the neural network, use of random dropout in the network, and the creation of initial perturbations with singular vector decomposition. The latter method is widely used in numerical weather prediction models, but is yet to be tested on neural networks. The ensemble mean forecasts obtained from these four approaches all beat the unperturbed neural network forecasts, with the retraining method yielding the highest improvement. However, the skill of the neural network forecasts is systematically lower than that of state‐of‐the‐art numerical weather prediction models.

Detection of Hepatoma based on Gene Expression using Unitary Matrix of Singular Vector Decomposition

Hepatoma is a long-term disease with a high risk of mortality. However, the disease is late detected, at the fourth level stadium due to silent symptoms. The infected hepatitis B virus gene HBx is a genome virus to trigger liver disease. This virus inserts material genetic into the host and disturbs the cell cycle. The regulation of gene expression is blocked to make work abnormal, especially for repairing and degrading. A microarray is a tool to quantify the RNA gene expression in huge volumes without any information for the related potential gene. Therefore, this study is proposed a feature extraction method using a unitary singular matrix for simplifying the classification model of hepatoma detection. Principally, the feature is decomposed using a singular vector to get the k-rank value of pattern. This matrix is applied to the representative machine learning algorithm, including KNN, Naive Bayes, C5.0 Decision Tree, and SVM. The experimental result achieved high performance with Area under the Curve (AUC) of above 90% on average.

Reconstruction of time series leaf area index for improving wheat yield estimates at field scales by fusion of Sentinel-2, -3 and MODIS imagery

Continuous time series crop growth monitoring during the main crop growth and development period at field scales is very important for crop management and yield estimation. For more than a decade, the time series leaf area index (LAI) products obtained from high temporal resolution satellites have been widely used in global crop growth monitoring. However, the spatial resolutions (250–1000 m) of these satellite sensors are too coarse for areas with complex and diverse land-use types, especially in China, which causes great uncertainties in crop growth monitoring and yield estimation results. In addition, due to the influence of clouds, optical remote sensing satellites cannot obtain continuous time series data at a given time step over the main crop growth and development period. In this paper, a method based on spatiotemporal data fusion and singular vector decomposition (SVD) is proposed to reconstruct field-scale time series LAI imagery over the main growth and development period of winter wheat. In this method, the Enhanced Spatial and Temporal Adaptive Reflectance Fusion Model (ESTARFM) is used to fuse the reflectance imagery of Sentinel-2 and Sentinel-3, and a linear regression model between the LAI data retrieved from the fused reflectance data and the singular vectors derived from the 4-day interval Moderate Resolution Imaging Spectroradiometer (MODIS) LAI data is established to reconstruct the continuous time series field-scale LAI imagery at a given time step. The accuracy of the reconstructed LAI and its capability for winter wheat yield estimation were tested on the Guanzhong Plain of China. The results indicate that (1) the ESTARFM model can fuse the reflectance bands from visible to shortwave infrared of Sentinel-2 and Sentinel-3 on the Guanzhong Plain accurately within a 20-day interval of the winter wheat growth and development period; (2) the 4-day interval field-scale LAI imagery over the main winter wheat growth and development period can be accurately reconstructed based on the linear regression models between the fused LAI data and the singular vectors derived from the 4-day interval MODIS LAI data; and (3) the yield map estimated from the reconstructed field-scale LAI shows more yield distribution details than MODIS yield estimation results. This study shows the feasibility of reconstructing continuous time series field-scale LAI data over the main winter wheat growth and development period on the Guanzhong Plain by combining the spatiotemporal data fusion model with SVD and the potential for estimating the winter wheat yield at field scales.