Search

To address the background-bias problem in computer-aided cytology caused by microscopic slide deterioration, this article proposes a deep learning approach for cell segmentation and background removal without requiring cell annotation. A U-Net-based model was trained to separate cells from the background in an unsupervised manner by leveraging the redundancy of the background and the sparsity of cells in liquid-based cytology (LBC) images. The experimental results demonstrate that the U-Net-based model trained on a small set of cytology images can exclude background features and accurately segment cells. This capability is beneficial for debiasing in the detection and classification of the cells of interest in oral LBC. Slide deterioration can significantly affect deep learning-based cell classification. Our proposed method effectively removes background features at no cost of cell annotation, thereby enabling accurate cytological diagnosis through the deep learning of microscopic slide images. [ABSTRACT FROM AUTHOR]

Hyperspectral target detection has a wide range of applications in marine target monitoring. Traditional methods for target detection take less consideration of the inherent structural information of hyperspectral images and make insufficient use of spatial information. These algorithms may experience degradation in efficacy during complex scenarios. To address these issues, this study introduces a hyperspectral target detection approach based on tensor adaptive reconstruction cascade spatial-spectral fusion, named as TRSSF. First, the position of the pixel that best matches the prior spectrum is obtained. Second, tensor decomposition and reconstruction of the original hyperspectral data are performed. Linear total variation smoothing is used to acquire the principal components in the spatial dimensionality unfolding of data, and correlation regularization robust principal component analysis is employed to derive the spectral dimensionality unfolding's principal components of data. Finally, the spatial-spectral fusion method is proposed for detecting hyperspectral targets on the reconstructed data. The use of multi-morphological feature fusion can fully utilize the spatial features to complement the spectral detection results and improve the integrity of target detection. The experiments conducted on the publicly available dataset and collected datasets demonstrated the effective detection achieved by the proposed method. [ABSTRACT FROM AUTHOR]

Lithium-ion batteries, with their high energy density, long cycle life, and non-polluting advantages, are widely used in energy storage stations. Connecting lithium batteries in series to form a battery pack can achieve the required capacity and voltage. However, as the batteries are used for extended periods, some individual cells in the battery pack may experience abnormal failures, affecting the performance and safety of the battery pack. At the same time, as batteries operate in complex environments, the data collected by sensors are susceptible to random noise and drift interference, which can affect the accuracy of anomaly detection in individual battery cells. In order to solve this problem, this article proposes an anomaly detection method for battery cells based on Robust Principal Component Analysis (RPCA), taking the historical operation and maintenance data of a large-scale battery pack from an energy storage station as the research subject. Firstly, theRPCA is used to denoise the observed voltage data of the battery cells to an extreme degree, obtaining a baseline charging state curve for a cell consistency assessment. This also solves the problem of sensor outputs being affected by random noise. To further detect and identify abnormal battery cells, the RPCA is used to extract outlier components. Based on the Average Deviation-3σ principle and by utilizing Gaussian distribution probability characteristics, battery cells are conducted to screen, and the serial numbers of the anomaly cells are obtained. Finally, the effectiveness and accuracy of this anomaly detection method for battery cells are compared and verified through different statistical distributions. [ABSTRACT FROM AUTHOR]

Robust principal component analysis (RPCA) is devoted to tackling grossly corrupted datasets with noise. However, the performance of RPCA is usually circumscribed by the lack of efficiency of singular value decomposition (SVD), which rules out its potential applications to many large-scale real-world problems. In this paper, we develop a nonconvex optimization algorithm customized to SVD-free RPCA models. The proposed algorithm, which is built upon proximal alternating linearized minimization Bolte et al. [Proximal alternating linearized minimization for nonconvex and nonsmooth problems. Math Program. 2014;146(1–2):459–494], can reduce computational efforts by partially linearizing data fidelity and increase efficiency by leveraging inertial techniques. Under the Kurdyka-Łojasiewicz assumption on the objective function and some mild premises on stepsizes, the sequence produced by the proposed algorithm converges globally to a critical point of SVD-free RPCA models. Numerical simulations on synthetic and real datasets demonstrate the compelling performance of the proposed algorithm. [ABSTRACT FROM AUTHOR]

Numerous studies have demonstrated the potential of robust principal component analysis (RPCA) in image recovery. However, conventional RPCA methods for color image recovery apply RPCA independently to each color channel, which ignores the correlation information of the red, green and blue channels. To improve the performance of RPCA-based methods and draw inspiration from the success of quaternion representation in color image processing, we propose an improved quaternion RPCA (IQRPCA) method for color image recovery. The IQRPCA method treats all color channels holistically and considers the correlation information of different color channels naturally. In addition, we have developed a quaternion nuclear norm known as improved quaternion Cauchy nuclear norm, which produces an even more effective and robust approach to color image recovery task. Compared to the RPCA method in the quaternion setting, the IQRPCA method treats the singular values differently, which shows better recovery performance than competing methods. Furthermore, we provide a convergence analysis of the proposed method. Our experiments confirm the effectiveness of IQRPCA in the application of color image recovery. [ABSTRACT FROM AUTHOR]

The received signal may suffer from continuous missing data due to reasons such as antenna damage, which affects the performance of the direction-of-arrival (DOA) estimation. In order to achieve a better DOA estimation performance, this paper presents a matrix recovery DOA estimation method based on the truncated nuclear norm (TrNN) robust principal component analysis (RPCA). Firstly, the received signal recovery and noise reducing problem of a single snapshot is modeled as a low-rank matrix RPCA recovery problem, and the F-norm is used to constrain the noise entries so as to better separate the low-rank received signal from the noise. Secondly, TrNN is introduced into the low-rank matrix RPCA model in order to make better use of the a priori information on the number of signal sources in the received signal. In the process of TrNN optimization, only large singular values are used corresponding to the number of sources for processing, which reduces the computational complexity of the nuclear norm optimization to some extent without losing the target signal information. The proposed algorithm is used to complete the data of all snapshots. Finally, the DOA parameters are estimated by means of subspace theory. Simulation results show that the proposed algorithm achieves better angular parameter estimation performance with associated low-rank matrix recovery at low SNR. [ABSTRACT FROM AUTHOR]

Robust principal component analysis (RPCA) is a widely used method for recovering low‐rank structure from data matrices corrupted by significant and sparse outliers. These corruptions may arise from occlusions, malicious tampering, or other causes for anomalies, and the joint identification of such corruptions with low‐rank background is critical for process monitoring and diagnosis. However, existing RPCA methods and their extensions largely do not account for the underlying probabilistic distribution for the data matrices, which in many applications are known and can be highly non‐Gaussian. We thus propose a new method called RPCA for exponential family distributions (eRPCA$$ {e}^{\mathrm{RPCA}} $$), which can perform the desired decomposition into low‐rank and sparse matrices when such a distribution falls within the exponential family. We present a novel alternating direction method of multiplier optimization algorithm for efficient eRPCA$$ {e}^{\mathrm{RPCA}} $$ decomposition, under either its natural or canonical parametrization. The effectiveness of eRPCA$$ {e}^{\mathrm{RPCA}} $$ is then demonstrated in two applications: the first for steel sheet defect detection and the second for crime activity monitoring in the Atlanta metropolitan area. [ABSTRACT FROM AUTHOR]

Studying the crack propagation of ultrahigh-performance concrete (UHPC) helps us understand its mechanical mechanism and assess its structural performance. A novel method for crack separation and its characteristic evaluation was developed in this work. The proposed method introduces robust principal component analysis (RPCA) to decompose a data matrix from video streams stacked into a low-rank matrix and a sparse matrix, in which the sparse matrix represents the crack information. Compared with the cracks in a binary image, the obtained sparse matrix preserves rich crack information that can be used to quantify crack characteristics. The statistical characteristics of the crack area, the major and minor axes of the equivalent ellipse for crack regions, and the power spectral density are investigated and compared continuously. The proposed method is demonstrated by the crack development of UHPC under tensile loading. The analysis results indicate that RPCA can accurately separate cracks from the background. In the frequency domain by performing the Fourier transform of the sparse matrix, cracks are concentrated at small wavenumbers and the magnitude of small wavenumbers increases with an increase in the crack width. The relationship between the crack propagation and the stress–strain is also discussed. This work provides insight into the crack propagation of UHPC and an accumulated crack database for predicting the damage evolution of UHPC. [ABSTRACT FROM AUTHOR]

In recent years, high-dimensional data, such as waveform signals and images have become ubiquitous. This type of data is often represented by multiway arrays or tensors. Several statistical models, including tensor regression, have been developed for such tensor data. However, these models are sensitive to the presence of arbitrary outliers within the tensors. To address the issue, this article proposes a Robust Tensor-On-Tensor (RTOT) regression approach, which has the capability of modeling high-dimensional data when the data is corrupted by outliers. Through several simulations and case studies, we evaluate the performance of the proposed method. The results reveal the advantage of the RTOT over some benchmarks in the literature in terms of estimation error. A Python implementation is available at . [ABSTRACT FROM AUTHOR]

Lithium-ion batteries, with their high energy density, long cycle life, and non-polluting advantages, are widely used in energy storage stations. Connecting lithium batteries in series to form a battery pack can achieve the required capacity and voltage. However, as the batteries are used for extended periods, some individual cells in the battery pack may experience abnormal failures, affecting the performance and safety of the battery pack. At the same time, as batteries operate in complex environments, the data collected by sensors are susceptible to random noise and drift interference, which can affect the accuracy of anomaly detection in individual battery cells. In order to solve this problem, this article proposes an anomaly detection method for battery cells based on Robust Principal Component Analysis (RPCA), taking the historical operation and maintenance data of a large-scale battery pack from an energy storage station as the research subject. Firstly, theRPCA is used to denoise the observed voltage data of the battery cells to an extreme degree, obtaining a baseline charging state curve for a cell consistency assessment. This also solves the problem of sensor outputs being affected by random noise. To further detect and identify abnormal battery cells, the RPCA is used to extract outlier components. Based on the Average Deviation-3σ principle and by utilizing Gaussian distribution probability characteristics, battery cells are conducted to screen, and the serial numbers of the anomaly cells are obtained. Finally, the effectiveness and accuracy of this anomaly detection method for battery cells are compared and verified through different statistical distributions.

CRISPR‐Cas9 screens facilitate the discovery of gene functional relationships and phenotype‐specific dependencies. The Cancer Dependency Map (DepMap) is the largest compendium of whole‐genome CRISPR screens aimed at identifying cancer‐specific genetic dependencies across human cell lines. A mitochondria‐associated bias has been previously reported to mask signals for genes involved in other functions, and thus, methods for normalizing this dominant signal to improve co‐essentiality networks are of interest. In this study, we explore three unsupervised dimensionality reduction methods—autoencoders, robust, and classical principal component analyses (PCA)—for normalizing the DepMap to improve functional networks extracted from these data. We propose a novel "onion" normalization technique to combine several normalized data layers into a single network. Benchmarking analyses reveal that robust PCA combined with onion normalization outperforms existing methods for normalizing the DepMap. Our work demonstrates the value of removing low‐dimensional signals from the DepMap before constructing functional gene networks and provides generalizable dimensionality reduction‐based normalization tools. Synopsis: Dimensionality reduction‐based methods are proposed for normalizing Cancer Dependency Map (DepMap) genome‐wide CRISPR screen data to enhance the functional information in co‐essentiality networks extracted from DepMap.Low‐dimensional patterns introduce dominant covariation in gene networks derived from DepMap data, obscuring more subtle functional relationships.Applying dimensionality reduction approaches to remove low‐dimensional signal, including robust and classical principal component analysis or autoencoders, can increase functional information captured by similarity networks derived from DepMap data.Onion normalization, which integrates several normalized data layers into a single network, outperforms existing methods for constructing co‐essentiality networks from the DepMap. [ABSTRACT FROM AUTHOR]

An increasing number of data science and machine learning problems rely on computation with tensors, which better capture the multi-way relationships and interactions of data than matrices. When tapping into this critical advantage, a key challenge is to develop computationally efficient and provably correct algorithms for extracting useful information from tensor data that are simultaneously robust to corruptions and ill-conditioning. This paper tackles tensor robust principal component analysis (RPCA), which aims to recover a low-rank tensor from its observations contaminated by sparse corruptions, under the Tucker decomposition. To minimize the computation and memory footprints, we propose to directly recover the low-dimensional tensor factors—starting from a tailored spectral initialization—via scaled gradient descent (ScaledGD), coupled with an iteration-varying thresholding operation to adaptively remove the impact of corruptions. Theoretically, we establish that the proposed algorithm converges linearly to the true low-rank tensor at a constant rate that is independent with its condition number, as long as the level of corruptions is not too large. Empirically, we demonstrate that the proposed algorithm achieves better and more scalable performance than state-of-the-art tensor RPCA algorithms through synthetic experiments and real-world applications. [ABSTRACT FROM AUTHOR]

Dysregulation of microRNAs (miRNAs) is closely associated with refractory human diseases, and the identification of potential associations between small molecule (SM) drugs and miRNAs can provide valuable insights for clinical treatment. Existing computational techniques for inferring potential associations suffer from limitations in terms of accuracy and efficiency. To address these challenges, we devise a novel predictive model called RPCA |$\Gamma $| NR, in which we propose a new Robust principal component analysis (PCA) framework based on |$\gamma $| -norm and |$l_{2,1}$| -norm regularization and design an Augmented Lagrange Multiplier method to optimize it, thereby deriving the association scores. The Gaussian Interaction Profile Kernel Similarity is calculated to capture the similarity information of SMs and miRNAs in known associations. Through extensive evaluation, including Cross Validation Experiments, Independent Validation Experiment, Efficiency Analysis, Ablation Experiment, Matrix Sparsity Analysis, and Case Studies, RPCA |$\Gamma $| NR outperforms state-of-the-art models concerning accuracy, efficiency and robustness. In conclusion, RPCA |$\Gamma $| NR can significantly streamline the process of determining SM–miRNA associations, thus contributing to advancements in drug development and disease treatment. [ABSTRACT FROM AUTHOR]

Abstract CRISPR‐Cas9 screens facilitate the discovery of gene functional relationships and phenotype‐specific dependencies. The Cancer Dependency Map (DepMap) is the largest compendium of whole‐genome CRISPR screens aimed at identifying cancer‐specific genetic dependencies across human cell lines. A mitochondria‐associated bias has been previously reported to mask signals for genes involved in other functions, and thus, methods for normalizing this dominant signal to improve co‐essentiality networks are of interest. In this study, we explore three unsupervised dimensionality reduction methods—autoencoders, robust, and classical principal component analyses (PCA)—for normalizing the DepMap to improve functional networks extracted from these data. We propose a novel “onion” normalization technique to combine several normalized data layers into a single network. Benchmarking analyses reveal that robust PCA combined with onion normalization outperforms existing methods for normalizing the DepMap. Our work demonstrates the value of removing low‐dimensional signals from the DepMap before constructing functional gene networks and provides generalizable dimensionality reduction‐based normalization tools.

Face masks are widely used in various industries and jobs, such as healthcare, food service, construction, manufacturing, retail, hospitality, transportation, education, and public safety. Masked face recognition is essential to accurately identify and authenticate individuals wearing masks. Masked face recognition has emerged as a vital technology to address this problem and enable accurate identification and authentication in masked scenarios. In this paper, we propose a novel method that utilizes a combination of deep-learning-based mask detection, landmark and oval face detection, and robust principal component analysis (RPCA) for masked face recognition. Specifically, we use pretrained ssd-MobileNetV2 for detecting the presence and location of masks on a face and employ landmark and oval face detection to identify key facial features. The proposed method also utilizes RPCA to separate occluded and non-occluded components of an image, making it more reliable in identifying faces with masks. To optimize the performance of our proposed method, we use particle swarm optimization (PSO) to optimize both the KNN features and the number of k for KNN. Experimental results demonstrate that our proposed method outperforms existing methods in terms of accuracy and robustness to occlusion. Our proposed method achieves a recognition rate of 97%, which is significantly higher than the state-of-the-art methods. Our proposed method represents a significant improvement over existing methods for masked face recognition, providing high accuracy and robustness to occlusion. [ABSTRACT FROM AUTHOR]

Motion capture (mocap) data is often corrupted by the presence of erroneous entries. Corrupted mocap data tends to deteriorate the structure of the recorded human actions, causing hindrance while using it for different applications. In this paper, we propose a new four-way optimization model for recovering plausible human actions from the misrepresented mocap sequences without the requirement of any pre-trained model. The proposed model employs joint ℓ 1 2 regularized low rank and sparse priors for separating the clean mocap data and the erroneous entries effectively. The brilliance of this work is confined in the efficient formulation of the optimization model by integrating two additional constraints based on the movement of individual node trajectories and the hierarchy of the parent-child node pairs. The former ensures the smoothness of the action sequences, while the latter limits the spatial drifting of the skeletal nodes. Detailed analysis of the experimental results showcases the ability of the developed algorithm in achieving commendable human action recovery in minimal time compared to the recent counterparts, even from mocap data with 20–30% of misrepresented joints in 90% of the frames. [ABSTRACT FROM AUTHOR]

Moving target detection (MTD) is a crucial task in computer vision applications. In this paper, we investigate the problem of detecting moving targets in infrared (IR) surveillance video sequences captured using a steady camera in a maritime setting. For this purpose, we employ robust principal component analysis (RPCA), which is an improvement of principal component analysis (PCA) that separates an input matrix into the following two matrices: a low-rank matrix that is representative, in our case study, of the slowly changing background, and a sparse matrix that is representative of the foreground. RPCA is usually implemented in a non-causal batch form. To pursue a real-time application, we tested an online implementation, which, unfortunately, was affected by the presence of the target in the scene during the initialization phase. Therefore, we improved the robustness by implementing a saliency-based strategy. The advantages offered by the resulting technique, which we called "saliency-aided online moving window RPCA" (S-OMW-RPCA) are the following: RPCA is implemented online; along with the temporal features exploited by RPCA, the spatial features are also taken into consideration by using a saliency filter; the results are robust against the condition of the scene during the initialization. Finally, we compare the performance of the proposed technique in terms of precision, recall, and execution time with that of an online RPCA, thus, showing the effectiveness of the saliency-based approach. [ABSTRACT FROM AUTHOR]

Studying and understanding the complexity and interactions of different factors influencing stream sediment quality is necessary for the development of successful water quality management strategies. This study aims to evaluate the level of contamination by potentially toxic elements (PTEs) (As, Co, Cr, Cu, Mn, Ni, Pb, V, Zn) of the stream sediments of the Nile River. During the spring of 2019, river sediments were sampled at 23 sites along the Nile River. For each sample, one aliquot was digested in aqua regia and analyzed by ICP-MS for pseudo-total concentration, while for another aliquot, sequential extraction procedures were applied to determine chemical speciation. Compositional data analysis (CoDa) and k-means were applied to recognize the contribution of natural and anthropogenic sources, while pollution indices (EF, RAC) and sediment quality guidelines (SQGs) were applied to assess the ecological risk to biotic species. The results reveal that elements such as Cr, Mn, V and Fe, found in high concentrations in almost all samples (Cr up to 739 mg/kg, Mn up to 1942 mg/kg, V up to 507 mg/kg, Fe up to 98,519 mg/kg), have a natural origin, while the concentrations of Cu (up to 69 mg/kg), Ni (up to 88 mg/kg), Co (up to 42 mg/kg) and As (up to 9.8 mg/kg) are linked to both natural and anthropogenic processes. Sequential extraction shows that Mn, Co, Ni and, in some sites, Cu and Zn, are the most bioavailable elements. These elements present a high risk of toxicity, while the remaining elements imply a low-to-moderate risk. [ABSTRACT FROM AUTHOR]

The environmental geochemical characterization of mineralized areas prior to mining does not receive adequate attention. This study shows trace element distribution in soils of two unexploited porphyry copper deposits located in Darreh-Zereshk and Ali-Abad in central Iran. The study was carried out using a compositional data analysis (CoDa) approach and combination of multivariate statistics and clustering techniques, which made it possible to identify the geochemical associations representing the different areas of the mineral deposits. The results of the chemical analyses, performed by ICP-MS, revealed high concentrations of those elements typically associated with porphyry deposits (As, Co, Cu, Mo, Ni, Pb, and Zn). The typical zonal pattern with an anomaly of Cu in central parts of the system and the prevalence of epithermal elements (Ag, Cd, Pb, and Zn) toward the peripheral propylitic alteration zone were recognized. The XRD analysis of selected soil samples allowed us to determine the distribution of elements within the different carrier minerals. Afterward, geochemical speciation patterns were investigated by a four-step sequential extraction procedure based on BCR protocol. The residual fraction consisting of primary resistant minerals was found to be the main host for As (73–93.4%), Cr (65.1–79.6%), Cu (54.3–81.4%), Ni (58.9–80.6%), V (75.9–88%), and Zn (56.5–60.5%) in the studied soils. Even though these elements are not readily leachable, their behavior and distribution could be largely affected by the mining operation and consequent changes in the physicochemical properties of the soil. The soluble–exchangeable phase was only less than 15% of the total extractions for all elements, except for Cd. With respect to the mobility factor (MF), Cd was the most mobile element followed by Sb and Pb. The measured risk assessment code (RAC) presented the following risk order: Cd > Sb > Ni > Co > Pb > Cr > As > Zn > Cu > V. This study reveals that the acquisition of pre-mining geo-environmental data of trace elements is very important to establish pre-mining backgrounds and baselines for evaluating post-mining or post-reclamation geochemical signatures. [ABSTRACT FROM AUTHOR]

In real-world scenarios, images may be affected by additional noise during compression and transmission, which interferes with postprocessing such as image segmentation and feature extraction. Image noise can also be induced by environmental variables and imperfections in the imaging equipment. Robust principal component analysis (RPCA), one of the traditional approaches for denoising images, suffers from a failure to efficiently use the background's low-rank prior information, which lowers its effectiveness under complex noise backgrounds. In this paper, we propose a robust PCA method based on a nonconvex low-rank approximation and total variational regularization (TV) to model the image denoising problem in order to improve the denoising performance. Firstly, we use a nonconvex γ -norm to address the issue that the traditional nuclear norm penalizes large singular values excessively. The rank approximation is more accurate than the nuclear norm thanks to the elimination of matrix elements with substantial approximation errors to reduce the sparsity error. The method's robustness is improved by utilizing the low sensitivity of the γ -norm to outliers. Secondly, we use the l 1 -norm to increase the sparsity of the foreground noise. The TV norm is used to improve the smoothness of the graph structure in accordance with the sparsity of the image in the gradient domain. The denoising effectiveness of the model is increased by employing the alternating direction multiplier strategy to locate the global optimal solution. It is important to note that our method does not require any labeled images, and its unsupervised denoising principle enables the generalization of the method to different scenarios for application. Our method can perform denoising experiments on images with different types of noise. Extensive experiments show that our method can fully preserve the edge structure information of the image, preserve important features of the image, and maintain excellent visual effects in terms of brightness smoothing. [ABSTRACT FROM AUTHOR]

Detection of omics sample outliers is important for preventing erroneous biological conclusions, developing robust experimental protocols, and discovering rare biological states. Two recent publications describe robust algorithms for detecting transcriptomic sample outliers, but neither algorithm had been incorporated into a software tool for scientists. Here we describe Ensemble Methods for Outlier Detection (EnsMOD) which incorporates both algorithms. EnsMOD calculates how closely the quantitation variation follows a normal distribution, plots the density curves of each sample to visualize anomalies, performs hierarchical cluster analyses to calculate how closely the samples cluster with each other, and performs robust principal component analyses to statistically test if any sample is an outlier. The probabilistic threshold parameters can be easily adjusted to tighten or loosen the outlier detection stringency. EnsMOD can be used to analyze any omics dataset with normally distributed variance. Here it was used to analyze a simulated proteomics dataset, a multiomic (proteome and transcriptome) dataset, a single-cell proteomics dataset, and a phosphoproteomics dataset. EnsMOD successfully identified all of the simulated outliers, and subsequent removal of a detected outlier improved data quality for downstream statistical analyses. [ABSTRACT FROM AUTHOR]

In previous works, the authors have shown the feasibility of using classical and robust principal component analysis for damage detection on structures when ultrasonic guided waves are used. It has also been demonstrated that robust principal component analysis presents a higher probability of detection accuracy when data are corrupted. In the present work, a robust principal component analysis orthogonal distance is proposed as a new feature for damage detection strategy based on ultrasonic guided waves on structures subjected to uniform temperature changes. The effect of this temperature fluctuation on the signal propagation and also in the new feature is analyzed. Temperature compensation is applied to mitigate the effect of temperature changes on the reliability of the damage detection methodology. The proposed feature and damage detection strategy that considers these effects are tested on two structures: a laboratory scale composite plate and a large-scale complex composite that is representative of a component from an aerospace application. The promising result proves the ability of the new feature as a damage detection tool.

In various computer vision applications, the moving object detection is an essential step. Principal Component Analysis (PCA) techniques are often used for this purpose. However, the performance of this method is degraded by camera shake, hidden moving objects, dynamic background scenes, and / or fluctuating exposure. Robust Principal Component Analysis (RPCA) is a useful approach for reducing stationary background noise as it can recover low rank matrices. That is, moving object is formed by the low power models and the static background of RPCA. This paper proposes a simple alternative minimization algorithm to fix minor discrepancies in the original Principal Component Pursuit (PCP) or RPCA function. A novel hybrid method of cartoon texture features used as a data matrix for RPCA taking into account low-ranking and rare matrix is presented. A new non-convex function is proposed to better control the low-range properties of the video background. Simulation results demonstrate that the proposed algorithm is capable of giving consistent random estimates and can indeed improve the accuracy of object recognition in comparison with existing methods.

The Heilongjiang Jianbiannongchang area is located at the confluence of the Great and Lesser Xing'an Ranges. This area has a complex magmatic and tectonic evolutionary history that has resulted in a complex and diverse geological background for mineralization. In this study, isometric logarithmic ratio (ILR) transformations of Au, Cu, Pb, Zn, and Sb contents were performed in the 1:50,000 soil geochemical data of the Jianbiannongchang area. Robust principal component analysis (RPCA) was conducted based on ILR transformation. The local singularity and spectrum-area (S-A) methods were used to extract information on mineralogic anomalies. The results showed that: (1) the transformed data eliminated the influence of the original data closure effect, and the PC1 and PC2 information obtained by applying RPCA reflected ore-producing element anomalies dominated by Au and Cu. (2) The local singularity method can enhance the information of the local strong and weak slow anomalies. After performing local singularity analysis on PC1 and PC2, the obtained local anomalies reflected the local singularity spatial anomaly patterns related to Cu and Au mineralization in this area, which is an effective method for trapping ore-producing anomalies. (3) Furthermore, the composite anomaly decomposition of PC1 and PC2 was performed using the S-A method, and the screened anomalous and background fields reflect the ore-producing anomalies related to Cu and Au mineralization. This information is in agreement with known Cu and Au mineralization. (4) The geochemical anomalies with mineralization potential were obtained outside the known mineralization sites by integrating the information of ore-producing anomalies extracted by the local singularity and S-A methods, providing the theoretical basis and exploration direction for future exploration in the study area. [ABSTRACT FROM AUTHOR]

Singing-voice separation is a separation task that involves a singing voice and musical accompaniment. In this paper, we propose a novel, unsupervised methodology for extracting a singing voice from the background in a musical mixture. This method is a modification of robust principal component analysis (RPCA) that separates a singing voice by using weighting based on gammatone filterbank and vocal activity detection. Although RPCA is a helpful method for separating voices from the music mixture, it fails when one single value, such as drums, is much larger than others (e.g., the accompanying instruments). As a result, the proposed approach takes advantage of varying values between low-rank (background) and sparse matrices (singing voice). Additionally, we propose an expanded RPCA on the cochleagram by utilizing coalescent masking on the gammatone. Finally, we utilize vocal activity detection to enhance the separation outcomes by eliminating the lingering music signal. Evaluation results reveal that the proposed approach provides superior separation outcomes than RPCA on ccMixter and DSD100 datasets. [ABSTRACT FROM AUTHOR]

Robust Principal Component Analysis (RPCA) offers a powerful tool for recovering a low-rank matrix from highly corrupted data, with growing applications in computational biology. Biological processes commonly form intrinsic hierarchical structures, such as tree structures of cell development trajectories and tumor evolutionary history. The rapid development of single-cell sequencing (SCS) technology calls for the recovery of embedded tree structures from noisy and heterogeneous SCS data. In this study, we propose RobustTree, a unified framework to reconstruct the inherent topological structure underlying highdimensional data with noise. By extending RPCA to handle tree structure optimization, RobustTree leverages data denoising, clustering, and tree structure reconstruction. It solves the tree optimization problem with an adaptive parameter selection scheme that we proposed. In addition to recovering real datasets, RobustTree can reconstruct continuous topological structure and discrete-state topological structure of underlying SCS data. We apply RobustTree on multiple synthetic and real datasets and demonstrate its high accuracy and robustness when analyzing high-noise SCS data with embedded complex structures. The code is available at https://github.com/ucasdp/ RobustTree. [ABSTRACT FROM AUTHOR]

Ground-penetrating radar (GPR) is an important nondestructive testing (NDT) tool for the underground exploration of urban roads. However, due to the large amount of GPR data, traditional manual interpretation is time-consuming and laborious. To address this problem, an efficient underground target detection method for urban roads based on neural networks is proposed in this paper. First, robust principal component analysis (RPCA) is used to suppress the clutter in the B-scan image. Then, three time-domain statistics of each A-scan signal are calculated as its features, and one backpropagation (BP) neural network is adopted to recognize A-scan signals to obtain the horizontal regions of targets. Next, the fusion and deletion (FAD) algorithm is used to further optimize the horizontal regions of targets. Finally, three time-domain statistics of each segmented A-scan signal in the horizontal regions of targets are extracted as the features, and another BP neural network is employed to recognize the segmented A-scan signals to obtain the vertical regions of targets. The proposed method is verified with both simulation and real GPR data. The experimental results show that the proposed method can effectively locate the horizontal ranges and vertical depths of underground targets for urban roads and has higher recognition accuracy and less processing time than the traditional segmentation recognition methods. [ABSTRACT FROM AUTHOR]

Moving object detection is one of the significant tasks in computer vision. Robust Principal Component Analysis (RPCA) is a common method for moving object detection. However, this algorithm fails to effectively utilize the low-rank prior information of the background and the spatio-temporal continuity of the foreground, and exhibits degraded performance in the existence of shadows, photometric variations, rapidly moving objects, dynamic backgrounds, and jitters. Therefore, a new model via RPCA framework using non-convex low-rank approximation and total variational regularization is proposed to detect moving objects. Firstly, this paper introduces a non-convex function to deal with the problem that the nuclear norm excessively penalizes large singular values for the sake of constraining the low-rank characteristics of video background availably. Then, the sparsity is strengthened with the l1-norm and the spatio-temporal continuity of the foreground is explored by TV regularization. Finally, the augmented Lagrange multiplier algorithm is expanded by the alternating direction multiplier strategy to solve the model. Extensive experiments show that the proposed method outperforms existing methods in terms of the accuracy of moving object detection and foreground extraction effect. [ABSTRACT FROM AUTHOR]

Freezing weather can introduce challenges in long-term structural health monitoring of civil structures, particularly bridges. A noticeable impact of freezing temperature is the emergence of sudden and sharp increases in structural modal frequencies, causing false alarm and mis-detection errors in change detection of civil structures. This paper proposes an innovative unsupervised data normalization method to mitigate freezing effects. The proposed method integrates locally robust principal component analysis (LRPCA) with Gaussian density distance (GDD) clustering, called GDD-LRPCA, which automatically determines the number of clusters. Initially, a training set of original modal frequencies is partitioned via the GDD clustering. Subsequently, an individual LRPCA model is fitted to each partition to extract new normalized modal frequencies insensitive to freezing effects. The groundbreaking nature of this research relies on developing an integrated unsupervised data normalizer by leveraging advanced machine learning algorithms such as local learning, robust learning, and hybrid unsupervised learning. The major advantage of the proposed method is its non-parametric nature obviating any supplementary technique for hyperparameter optimization. The validity of this method is benchmarked by real-world bridge structures along with several comparative analyses. Results demonstrate that GDD-LRPCA effectively removes the freezing effects from structural modal frequencies and outperforms its counterparts in unsupervised data normalization. • Proposing an unsupervised data normalizer for removing freezing effects from modal frequencies under statistical learning. • Leveraging cutting-edge machine learning algorithms including local learning, robust learning, and hybrid learning. • Suggesting a non-parametric framework for unsupervised data normalization without any hyperparameter optimization • Lacking the need for temperature sensor installation and measurement for removing freezing effects. [ABSTRACT FROM AUTHOR]

Deep multi-view subspace clustering has achieved increasing attention owing to its encouraging ability to address the nonlinear multi-view data. Despite significant progress, most existing deep multi-view subspace clustering methods still suffer from the following drawbacks. First, they usually focus on utilizing the last-level feature from the last encoder layer of the auto-encoder related to each view, yet often overlook exploiting the earlier-level features from the earlier encoder layers as well as integrating the multi-level features from multiple encoder layers. Second, when learning the subspace representation from the self-expressive layer, they rarely consider the cleanliness and robustness of the learned subspace representation. To overcome these drawbacks, this paper proposes a novel c lean and r obust m ulti-level s ubspace r epresentations l earning for deep multi-view subspace clustering (CRMSRL). Specifically, the auto-encoder is adopted to nonlinearly map each view into the multi-level features, and then multiple self-expressive layers are added between the encoder layers and their corresponding decoder layers to provide multiple information flow paths through the auto-encoder and learn the multi-level subspace representations corresponding to the multi-level features. Consequently, the intricate hierarchical information embedded in the multi-level features is passed to the corresponding multi-level subspace representations. Moreover, with the idea of robust principal component analysis (RPCA), the learned multi-level subspace representations can be further purified to achieve the cleaner and more robust ones. In addition, to excavate the structural consistency among different views, the common layer is designed to interact with different view-specific self-expressive layers and achieve the high-quality common subspace representation, which can be applied to the spectral clustering algorithm to obtain the final clustering results. Experimental results show that CRMSRL outperforms twelve state-of-the-art baseline methods on six benchmark datasets. In particular, CRMSRL achieves 4.5% improvements on the MSRCV1 dataset in terms of accuracy, compared with the best baseline method. • We propose a new method to handle the nonlinear multi-view subspace clustering task. • We use multiple self-express layers to learn multi-level subspace representations. • We use robust principal component analysis to clean up the learned representations. • We introduce the common layer to learn the shared common subspace representation. • Our method has better clustering performance than other state-of-the-art methods. [ABSTRACT FROM AUTHOR]

Abstract In this study, we propose a methodology for separating a singing voice from musical accompaniment in a monaural musical mixture. The proposed method uses robust principal component analysis (RPCA), followed by postprocessing, including median filter, morphology, and high-pass filter, to decompose the mixture. Subsequently, a deep recurrent neural network comprising two jointly optimized parallel-stacked recurrent neural networks (sRNNs) with mask layers and trained on limited data and computation is applied to the decomposed components to optimize the final estimated separated singing voice and background music to further correct misclassified or residual singing and background music in the initial separation. The experimental results of MIR-1K, ccMixter, and MUSDB18 datasets and the comparison with ten existing techniques indicate that the proposed method achieves competitive performance in monaural singing voice separation. On MUSDB18, the proposed method reaches the comparable separation quality in less training data and lower computational cost compared to the other state-of-the-art technique.

Rolling element bearings typically operate with fluctuating speed, leading to nonstationary vibrations. Moreover, bearings vibration signals are frequently hidden by strong distributions, making it difficult to detect clear bearing fault characteristics for diagnosis. Under this circumstance, the key issue is effectively extracting the transient features from the background interference and highlighting the time-varying fault characteristics. To address this issue, a sparse and low-rank decomposition approach is proposed. In this study, the sparsity of the variable defective characteristics and low-rank of background interference is revealed and exploited for bearing fault detection. Firstly, the time–frequency representation (TFR) of the envelope of measured signal is generated by the time–frequency transform. Then, a sparse and low-rank decomposition model is established based on robust principal component analysis (RPCA) to denoise the measured time–frequency representation and gain the sparse component. Finally, a time–frequency reassignment strategy is utilized to further enhance the capability of detecting the faulty characteristics in the decomposed sparse TFR. The synthetic and actual signals are evaluated to illustrate the reliability and efficacy of the proposed technique. The superiority is also validated by comparisons with STFT, synchrosqueezing transform (SST), ridge extraction method, and scaling-basis chirplet transform (SBCT). • A sparse and low-rank decomposition of time–frequency representation method is proposed for bearing fault diagnosis. • The sparsity of the time-varying fault characteristics and the low-rank property of noise are revealed for the first time. • The strong background interference can be well removed to highlight the fault features. • The effectiveness of the proposed method in simulated signal and actual situation is validated. [ABSTRACT FROM AUTHOR]

The Eastern Route of China's South-to-North Water Diversion Project (SNWDP-ER) traverses through impounded lakes that are potentially vulnerable to heavy metals (HMs) contamination although the understanding remains elusive. This study employed machine learning approaches, including super-clustering of Self-Organizing Map (SOM) and Robust Principal Component Analysis (RPCA), to elucidate the spatiotemporal patterns and assess ecological risks associated with HMs in the surface sediments of Gao-Bao-Shaobo Lake (GBSL) and Dongping Lake (DPL). We collected 184 surface sediments from 47 stations across the two important impounded lakes over four seasons. The results revealed higher HMs concentrations in the south-central GBSL and west-central DPL, with a notable increase in contamination in autumn. The comprehensive risk assessment, utilizing various indicators such as the Sediment Quality Guidelines (SQGs), Improved Potential Ecological Risk Index (IPERI), Geo-accumulation Index (Igeo), Contamination Factor (CF), and Enrichment Factor (EF), identified arsenic (As), cadmium (Cd), nickel (Ni), and chromium (Cr) as primary contaminants of concern. Positive Matrix Factorization (PMF) model, coupled with Spearman analysis, attributed over 70 % of HMs pollution to anthropogenic activities. This research provides a nuanced understanding of HMs pollution in the context of large-scale water diversion projects and offers a scientific basis for targeted pollution mitigation strategies., Competing Interests: Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper., (Copyright © 2024 Elsevier B.V. All rights reserved.)

Face masks are widely used in various industries and jobs, such as healthcare, food service, construction, manufacturing, retail, hospitality, transportation, education, and public safety. Masked face recognition is essential to accurately identify and authenticate individuals wearing masks. Masked face recognition has emerged as a vital technology to address this problem and enable accurate identification and authentication in masked scenarios. In this paper, we propose a novel method that utilizes a combination of deep-learning-based mask detection, landmark and oval face detection, and robust principal component analysis (RPCA) for masked face recognition. Specifically, we use pretrained ssd-MobileNetV2 for detecting the presence and location of masks on a face and employ landmark and oval face detection to identify key facial features. The proposed method also utilizes RPCA to separate occluded and non-occluded components of an image, making it more reliable in identifying faces with masks. To optimize the performance of our proposed method, we use particle swarm optimization (PSO) to optimize both the KNN features and the number of k for KNN. Experimental results demonstrate that our proposed method outperforms existing methods in terms of accuracy and robustness to occlusion. Our proposed method achieves a recognition rate of 97%, which is significantly higher than the state-of-the-art methods. Our proposed method represents a significant improvement over existing methods for masked face recognition, providing high accuracy and robustness to occlusion.

Studying and understanding the complexity and interactions of different factors influencing stream sediment quality is necessary for the development of successful water quality management strategies. This study aims to evaluate the level of contamination by potentially toxic elements (PTEs) (As, Co, Cr, Cu, Mn, Ni, Pb, V, Zn) of the stream sediments of the Nile River. During the spring of 2019, river sediments were sampled at 23 sites along the Nile River. For each sample, one aliquot was digested in aqua regia and analyzed by ICP-MS for pseudo-total concentration, while for another aliquot, sequential extraction procedures were applied to determine chemical speciation. Compositional data analysis (CoDa) and k-means were applied to recognize the contribution of natural and anthropogenic sources, while pollution indices (EF, RAC) and sediment quality guidelines (SQGs) were applied to assess the ecological risk to biotic species. The results reveal that elements such as Cr, Mn, V and Fe, found in high concentrations in almost all samples (Cr up to 739 mg/kg, Mn up to 1942 mg/kg, V up to 507 mg/kg, Fe up to 98,519 mg/kg), have a natural origin, while the concentrations of Cu (up to 69 mg/kg), Ni (up to 88 mg/kg), Co (up to 42 mg/kg) and As (up to 9.8 mg/kg) are linked to both natural and anthropogenic processes. Sequential extraction shows that Mn, Co, Ni and, in some sites, Cu and Zn, are the most bioavailable elements. These elements present a high risk of toxicity, while the remaining elements imply a low-to-moderate risk.

In real-world scenarios, images may be affected by additional noise during compression and transmission, which interferes with postprocessing such as image segmentation and feature extraction. Image noise can also be induced by environmental variables and imperfections in the imaging equipment. Robust principal component analysis (RPCA), one of the traditional approaches for denoising images, suffers from a failure to efficiently use the background’s low-rank prior information, which lowers its effectiveness under complex noise backgrounds. In this paper, we propose a robust PCA method based on a nonconvex low-rank approximation and total variational regularization (TV) to model the image denoising problem in order to improve the denoising performance. Firstly, we use a nonconvex γ-norm to address the issue that the traditional nuclear norm penalizes large singular values excessively. The rank approximation is more accurate than the nuclear norm thanks to the elimination of matrix elements with substantial approximation errors to reduce the sparsity error. The method’s robustness is improved by utilizing the low sensitivity of the γ-norm to outliers. Secondly, we use the l1-norm to increase the sparsity of the foreground noise. The TV norm is used to improve the smoothness of the graph structure in accordance with the sparsity of the image in the gradient domain. The denoising effectiveness of the model is increased by employing the alternating direction multiplier strategy to locate the global optimal solution. It is important to note that our method does not require any labeled images, and its unsupervised denoising principle enables the generalization of the method to different scenarios for application. Our method can perform denoising experiments on images with different types of noise. Extensive experiments show that our method can fully preserve the edge structure information of the image, preserve important features of the image, and maintain excellent visual effects in terms of brightness smoothing.

The kernel principal component analysis (KPCA) serves as an efficient approach for dimensionality reduction. However, the KPCA method is sensitive to the outliers since the large square errors tend to dominate the loss of KPCA. To strengthen the robustness of KPCA method, we propose a novel robust kernel principal component analysis with optimal mean (RKPCA-OM) method. RKPCA-OM not only possesses stronger robustness for outliers than the conventional KPCA method, but also can eliminate the optimal mean automatically. What is more, the theoretical proof proves the convergence of the algorithm to guarantee that the optimal subspaces and means are obtained. Lastly, exhaustive experimental results verify the superiority of our method. [ABSTRACT FROM AUTHOR]

The Heilongjiang Duobaoshan area is located at the confluence of the Great Xing'an Range and the Lesser Xing'an Range, and the area has undergone a complex magmatic and tectonic evolutionary history resulting in a complex and diverse geological background for mineralization. As a result of this geological complexity and the multi-period nature of mineralization, the geochemical data of the area are usually not satisfied with a single statistical distribution form, so traditional statistical methods cannot adequately explore and identify the distribution of deep-seated information in the geochemical data. Based on the above problems, this paper adopts a multivariate component data analysis method to process 14 mass fraction data elements, namely Ag, As, Au, Bi, Cu, Fe, Hg, Mn, Mo, Ni, Pb, Sb, W, and Zn, in the 1:50,000 soil geochemical data from the Duobaoshan area of Heilongjiang. The spatial distribution and internal structural characteristics of raw, logarithmic transformation and isometric logarithmic ratio (ILR) transformed data were compared using exploratory data analysis (EDA); robust principal component analysis (RPCA) was applied to obtain the PC1 and PC2 principal component combinations associated with mineralization, and a spectrum–area (S–A) fractal model was further used to decompose the geochemical anomalies of the PC1 and PC2 principal component combinations as composite anomalies. The results show the following: (i) The data transformed by the isometric logarithmic ratio (ILR) eliminate the influence of the original data closure effect, and the spatial scale of the data is more uniform; the data are approximately normally distributed, based on which RPCA can be applied to better explore the correlation between elements and the pattern of co-associated combinations. (ii) The S–A method was further used to decompose the composite anomalies of the PC1 and PC2 principal component combination in the study area. The anomalous and background fields of the screened-out PC1 and PC2 principal component combinations reflect anomalous information on mineralization dominated by Au mineralization. Moreover, the anomaly and background information after extraction were in good agreement with the known Au deposits (points), and many geochemical anomalies with prospecting potential were obtained in the periphery, providing a theoretical basis and exploration focus for the next step in the searching and exploring of the study area. [ABSTRACT FROM AUTHOR]

Extracting genes involved in cancer lesions from gene expression data is critical for cancer research and drug development. The method of feature selection has attracted much attention in the field of bioinformatics. Principal Component Analysis (PCA) is a widely used method for learning low-dimensional representation. Some variants of PCA have been proposed to improve the robustness and sparsity of the algorithm. However, the existing methods ignore the high-order relationships between data. In this paper, a new model named Robust Principal Component Analysis via Hypergraph Regularization (HRPCA) is proposed. In detail, HRPCA utilizes L2,1-norm to reduce the effect of outliers and make data sufficiently row-sparse. And the hypergraph regularization is introduced to consider the complex relationship among data. Important information hidden in the data are mined, and this method ensures the accuracy of the resulting data relationship information. Extensive experiments on multi-view biological data demonstrate that the feasible and effective of the proposed approach. [ABSTRACT FROM AUTHOR]

Audio–Visual Segmentation (AVS) aims to extract the sounding objects from a video. The current learning-based AVS methods are often supervised, which rely on specific task data annotations and expensive model training. Recognizing that the video background captured by a static camera is represented as a low-rank matrix, we introduce the non-convex robust principal component analysis into AVS task in this paper. This approach is unsupervised and only relies on input data patterns. Specifically, the proposed method decomposes each modality into the sum of two parts: the low-rank part that represents the background audio and visual information, and the sparse part that represents the foreground information. Furthermore, CUR decomposition is employed at each iteration to reduce the computational complexity in optimization. The experimental results also show that the proposed AVS outperforms the supervised methods on AVS-Bench Single-Source datasets. • Firstly introduces non-convex RPCA for Audio–Visual Segmentation. • Applies CUR decomposition to reduce computational complexity. • The task of unsupervised Audio–Visual Segmentation. • Demonstrates efficiency and effectiveness results on the AVS dataset. [ABSTRACT FROM AUTHOR]

Accurate and real-time separation of blood signal from clutter and noise signals is a critical step in clinical non-contrast ultrasound microvascular imaging. Despite the widespread adoption of singular value decomposition (SVD) and robust principal component analysis (RPCA) for clutter filtering and noise suppression, the SVD's sensitivity to threshold selection, along with the RPCA's limitations in undersampling conditions and heavy computational burden often result in suboptimal performance in complex clinical applications. To address those challenges, this study presents a novel low-rank prior-based fast RPCA (LP-fRPCA) approach to enhance the adaptability and robustness of clutter filtering and noise suppression with reduced computational cost. A low-rank prior constraint is integrated into the non-convex RPCA model to achieve a robust and efficient approximation of clutter subspace, while an accelerated alternating projection iterative algorithm is developed to improve convergence speed and computational efficiency. The performance of the LP-fRPCA method was evaluated against SVD with a tissue/blood threshold (SVD1), SVD with both tissue/blood and blood/noise thresholds (SVD2), and the classical RPCA based on the alternating direction method of multipliers algorithm through phantom and in vivo non-contrast experiments on rabbit kidneys. In the slow flow phantom experiment of 0.2 mm/s, LP-fRPCA achieved an average increase in contrast ratio (CR) of 10.68 dB, 9.37 dB, and 8.66 dB compared to SVD1, SVD2, and RPCA, respectively. In the in vivo rabbit kidney experiment, the power Doppler results demonstrate that the LP-fRPCA method achieved a superior balance in the trade-off between insufficient clutter filtering and excessive suppression of blood flow. Additionally, LP-fRPCA significantly reduced the runtime of RPCA by up to 94-fold. Consequently, the LP-fRPCA method promises to be a potential tool for clinical non-contrast ultrasound microvascular imaging. • LP-fRPCA is introduced for clutter filtering in ultrasound microvascular imaging. • Low-rank prior and accelerated algorithm is used for rapid convergence. • Superior microvascular visualization is demonstrated with enhanced CR and CNR. • Robust performance in both phantom and in vivo rabbit kidney experiments is shown. • Processing time is significantly reduced, improving clinical applicability. [ABSTRACT FROM AUTHOR]

Singing-voice separation is a separation task that involves a singing voice and musical accompaniment. In this paper, we propose a novel, unsupervised methodology for extracting a singing voice from the background in a musical mixture. This method is a modification of robust principal component analysis (RPCA) that separates a singing voice by using weighting based on gammatone filterbank and vocal activity detection. Although RPCA is a helpful method for separating voices from the music mixture, it fails when one single value, such as drums, is much larger than others (e.g., the accompanying instruments). As a result, the proposed approach takes advantage of varying values between low-rank (background) and sparse matrices (singing voice). Additionally, we propose an expanded RPCA on the cochleagram by utilizing coalescent masking on the gammatone. Finally, we utilize vocal activity detection to enhance the separation outcomes by eliminating the lingering music signal. Evaluation results reveal that the proposed approach provides superior separation outcomes than RPCA on ccMixter and DSD100 datasets.

A micro-expression is a subtle, local and brief facial movement. It can reveal the genuine emotions that a person tries to conceal and is considered an important clue for lie detection. The micro-expression research has attracted much attention due to its promising applications in various fields. However, due to the short duration and low intensity of micro-expression movements, micro-expression recognition faces great challenges, and the accuracy still demands improvement. To improve the efficiency of micro-expression feature extraction, inspired by the psychological study of attentional resource allocation for micro-expression cognition, we propose a deep local-holistic network method for micro-expression recognition. Our proposed algorithm consists of two sub-networks. The first is a Hierarchical Convolutional Recurrent Neural Network (HCRNN), which extracts the local and abundant spatio-temporal micro-expression features. The second is a Robust principal-component-analysis-based recurrent neural network (RPRNN), which extracts global and sparse features with micro-expression-specific representations. The extracted effective features are employed for micro-expression recognition through the fusion of sub-networks. We evaluate the proposed method on combined databases consisting of the four most commonly used databases, i.e., CASME, CASME II, CAS(ME) 2 , and SAMM. The experimental results show that our method achieves a reasonably good performance. [ABSTRACT FROM AUTHOR]