Showing papers on "Singular value decomposition published in 2022"

Geographical POI recommendation for Internet of Things: A federated learning approach using matrix factorization

Abstract: With the popularity of Internet of Things (IoT), Point-of-Interest (POI) recommendation has become an important application for location-based services (LBS). Meanwhile, there is an increasing requirement from IoT devices on the privacy of user sensitive data via wireless communications. In order to provide preferable POI recommendations while protecting user privacy of data communication in a distributed collaborative environment, this paper proposes a federated learning (FL) approach of geographical POI recommendation. The POI recommendation is formulated by an optimization problem of matrix factorization, and singular value decomposition (SVD) technique is applied for matrix decomposition. After proving the nonconvex property of the optimization problem, we further introduce stochastic gradient descent (SGD) into SVD and design an FL framework for solving the POI recommendation problem in a parallel manner. In our FL scheme, only calculated gradient information is uploaded from users to the FL server while all the users manage their rating and geographic preference data on their own devices for privacy protection during communications. Finally, real-world dataset from large-scale LBS enterprise is adopted for conducting extensive experiments, whose experimental results validate the efficacy of our approach.

Hyperspectral Image Denoising Using Factor Group Sparsity-Regularized Nonconvex Low-Rank Approximation

Abstract: Hyperspectral image (HSI) mixed noise removal is a fundamental problem and an important preprocessing step in remote sensing fields. The low-rank approximation-based methods have been verified effective to encode the global spectral correlation for HSI denoising. However, due to the large scale and complexity of real HSI, previous low-rank HSI denoising techniques encounter several problems, including coarse rank approximation (such as nuclear norm), the high computational cost of singular value decomposition (SVD) (such as Schatten $p$ -norm), and adaptive rank selection (such as low-rank factorization). In this article, two novel factor group sparsity-regularized nonconvex low-rank approximation (FGSLR) methods are introduced for HSI denoising, which can simultaneously overcome the mentioned issues of previous works. The FGSLR methods capture the spectral correlation via low-rank factorization, meanwhile utilizing factor group sparsity regularization to further enhance the low-rank property. It is SVD-free and robust to rank selection. Moreover, FGSLR is equivalent to Schatten $p$ -norm approximation ( Theorem 1 ), and thus FGSLR is tighter than the nuclear norm in terms of rank approximation. To preserve the spatial information of HSI in the denoising process, the total variation regularization is also incorporated into the proposed FGSLR models. Specifically, the proximal alternating minimization is designed to solve the proposed FGSLR models. Experimental results have demonstrated that the proposed FGSLR methods significantly outperform existing low-rank approximation-based HSI denoising methods.

RCEN: A Deep-Learning-Based Background Noise Suppression Method for DAS-VSP Records

Abstract: Recently, distributed optical fiber acoustic sensing (DAS) is regarded as a transformative technology in seismic exploration. However, both various complex background noise and weak desired signals significantly limit its practical application. To explore an effective denoising method for the vertical seismic profile (VSP) record received by DAS, we propose an improved residual encoder–decoder deep neural network (RED-Net) enhanced by deep iterative memory block (DMB) and channel aggregation block (CAB), called residual channel aggregation encoder–decoder network (RCEN). Here, DMB uses the weight accumulation theory to improve the feature extraction ability and achieve accurate noise elimination. Meanwhile, CAB, using the multi-channel analysis architecture, enhances the weak signal retention performance. In addition, we leverage both the synthetic data obtained by forward modeling and real DAS noise data to construct a sufficient training dataset with high authenticity, thereby meeting the requirement of network training. Both the synthetic and field DAS-VSP data processing results demonstrate the advantage of RCEN compared with competing algorithms, including singular value decomposition (SVD), conventional RED-Net, and feed-forward denoising convolutional neural network (DnCNN).

Low-Rank High-Order Tensor Completion With Applications in Visual Data

Abstract: Recently, tensor Singular Value Decomposition (t-SVD)-based low-rank tensor completion (LRTC) has achieved unprecedented success in addressing various pattern analysis issues. However, existing studies mostly focus on third-order tensors while order- d ( d ≥ 4 ) tensors are commonly encountered in real-world applications, like fourth-order color videos, fourth-order hyper-spectral videos, fifth-order light-field images, and sixth-order bidirectional texture functions. Aiming at addressing this critical issue, this paper establishes an order- d tensor recovery framework including the model, algorithm and theories by innovatively developing a novel algebraic foundation for order- d t-SVD, thereby achieving exact completion for any order- d low t-SVD rank tensors with missing values with an overwhelming probability. Emperical studies on synthetic data and real-world visual data illustrate that compared with other state-of-the-art recovery frameworks, the proposed one achieves highly competitive performance in terms of both qualitative and quantitative metrics. In particular, as the observed data density becomes low, i.e., about 10%, the proposed recovery framework is still significantly better than its peers. The code of our algorithm is released at https://github.com/Qinwenjinswu/TIP-Code.

Zeroing Neural Network for Pseudoinversion of an Arbitrary Time-Varying Matrix Based on Singular Value Decomposition

Abstract: Many researchers have investigated the time-varying (TV) matrix pseudoinverse problem in recent years, for its importance in addressing TV problems in science and engineering. In this paper, the problem of calculating the inverse or pseudoinverse of an arbitrary TV real matrix is considered and addressed using the singular value decomposition (SVD) and the zeroing neural network (ZNN) approaches. Since SVD is frequently used to compute the inverse or pseudoinverse of a matrix, this research proposes a new ZNN model based on the SVD method as well as the technique of Tikhonov regularization, for solving the problem in continuous time. Numerical experiments, involving the pseudoinversion of square, rectangular, singular, and nonsingular input matrices, indicate that the proposed models are effective for solving the problem of the inversion or pseudoinversion of time varying matrices.

Hybrid Nature-Inspired Optimization and Encryption-Based Watermarking for E-Healthcare

Abstract: With the growth and popularity of the utilization of medical images in smart healthcare, the security of these images using watermarks is one of the most recent research topics. This algorithm is based on the joint use of dual watermarking, nature-inspired optimization, and encryption schemes utilizing redundant-discrete wavelet transform (RDWT) and randomized-singular value decomposition (RSVD). The key idea of the proposed method is to embed system encoded media access control (MAC) address in patient's ID card image via discrete wavelet transform (DWT) to generate the final mark. Afterward, embed the generated watermark into computed tomography (CT) scan images of the COVID-19 patient and general images through employing the RDWT and RSVD. Further, we use a hybrid of particle swarm optimization (PSO) and Firefly optimization techniques to determine the optimal scaling factor for embedding purposes. After that, the watermarked CT scan image is encrypted using an encryption technique based on a nonlinear-chaotic map, random permutation, and singular value decomposition (SVD). Extensive evaluations establish the benefit of our proposed algorithm over the traditional schemes. The optimal robustness is more effective than the five traditional schemes at lower computational efficiency. IEEE

A truncated SVD-based ARIMA model for multiple QoS prediction in mobile edge computing

Abstract: In the mobile edge computing environments, Quality of Service (QoS) prediction plays a crucial role in web service recommendation. Because of distinct features of mobile edge computing, i.e., the mobility of users and incomplete historical QoS data, traditional QoS prediction approaches may obtain less accurate results in the mobile edge computing environments. In this paper, we treat the historical QoS values at different time slots as a temporal sequence of QoS matrices. By incorporating the compressed matrices extracted from QoS matrices through truncated Singular Value Decomposition (SVD) with the classical ARIMA model, we extend the ARIMA model to predict multiple QoS values simultaneously and efficiently. Experimental results show that our proposed approach outperforms the other state-of-the-art approaches in accuracy and efficiency.

Adaptive Event-Triggered Nonfragile State Estimation for Fractional-Order Complex Networked Systems With Cyber Attacks

Abstract: This article addresses the adaptive event-triggered nonfragile state estimation for the fractional-order complex networked systems subject to randomly occurring nonlinearities and adversarial network attacks, in which the order of fractional derivative operator satisfies $0 < p < 1$ . To reduce unnecessary transmission burden as much as possible in an allowable range, an adaptive event-triggered scheme (AETS) is introduced to determine whether the data released by the sensor should be transmitted to the nonfragile state estimator. First, based on the considerations of the designed AETS and stochastic cyber-attacks, one constructs a newly fractional-order estimation error system model. Then, by employing the Lyapunov functional approach and the properties of Mittag–Leffler (M–L) functions, a sufficient condition is obtained to ensure the augmented error system stochastic mean-square stability; moreover, by making use of matrix’s singular value decomposition (SVD), the desired nonfragile state estimator is designed, and the estimator gains can be obtained by finding the feasible solutions of linear matrix inequality (LMI). Finally, a numerical example and Chua’s circuit model example are given to illustrate the feasibility of the designed nonfragile estimator.

An FPP-resistant SVD-based image watermarking scheme based on chaotic control

Abstract: Image watermarking commonly involves singular value decomposition (SVD) because of its simplicity and minimal effect on image quality. However, SVD-based image watermarking schemes suffer from some drawbacks such as the false positive problem (FPP), and an undesirable trade-off between vital properties such as imperceptibility, embedding capacity, and robustness. To address these drawbacks, we improve upon the SVD-based color image watermarking by incorporating integer wavelet transform (IWT) and chaotic maps. A grayscale image watermark is decomposed into eight gray levels (bit-planes) before being encrypted and re-ordered by a chaotic sequence. Each encrypted bit-plane is inserted into singular values of the sub-bands of a host image. The embedding process is controlled by chaos-based multiple scaling factors (MSF) which are updated in each embedded bit-plane. The resulting values are involved in generating a hash value at the end of the embedding processes. The hash value is used to overcome FPP issues and improves security. Our findings illustrate the proposed scheme’s robustness, security, imperceptibility, and capacity. It also exhibits excellent robustness against attacks and its performance surpasses a variety of existing schemes. In addition, the proposed scheme is hypersensitive to even the slightest secret key variations.

Fast Tensor Nuclear Norm for Structured Low-Rank Visual Inpainting

Abstract: Low-rank modeling has achieved great success in visual data completion. However, the low-rank assumption of original visual data may be in approximate mode, which leads to suboptimality for the recovery of underlying details, especially when the missing rate is extremely high. In this paper, we go further by providing a detailed analysis about the rank distributions in Hankel structured and clustered cases, and figure out both non-local similarity and patch-based structuralization play a positive role. This motivates us to develop a new Hankel low-rank tensor recovery method that is competent to truthfully capture the underlying details with sacrifice of slightly more computational burden. First, benefiting from the correlation of different spectral bands and the smoothness of local spatial neighborhood, we divide the visual data into overlapping 3D patches and group the similar ones into individual clusters exploring the non-local similarity. Second, the 3D patches are individually mapped to the structured Hankel tensors for better revealing low-rank property of the image. Finally, we solve the tensor completion model via the well-known alternating direction method of multiplier (ADMM) optimization algorithm. Due to the fact that size expansion happens inevitably in Hankelization operation, we further propose a fast randomized skinny tensor singular value decomposition (rst-SVD) to accelerate the per-iteration running efficiency. Extensive experimental results on real world datasets verify the superiority of our method compared to the state-of-the-art visual inpainting approaches.

Multiview Subspace Clustering by an Enhanced Tensor Nuclear Norm

Abstract: Despite the promising preliminary results, tensor-singular value decomposition (t-SVD)-based multiview subspace is incapable of dealing with real problems, such as noise and illumination changes. The major reason is that tensor-nuclear norm minimization (TNNM) used in t-SVD regularizes each singular value equally, which does not make sense in matrix completion and coefficient matrix learning. In this case, the singular values represent different perspectives and should be treated differently. To well exploit the significant difference between singular values, we study the weighted tensor Schatten $p$ -norm based on t-SVD and develop an efficient algorithm to solve the weighted tensor Schatten $p$ -norm minimization (WTSNM) problem. After that, applying WTSNM to learn the coefficient matrix in multiview subspace clustering, we present a novel multiview clustering method by integrating coefficient matrix learning and spectral clustering into a unified framework. The learned coefficient matrix well exploits both the cluster structure and high-order information embedded in multiview views. The extensive experiments indicate the efficiency of our method in six metrics.

Inverse Radiative Heat Transfer

Abstract: This focus of this chapter is on inverse problems in radiative heat transfer. Following a brief discussion of the type of inverse problems encountered in this field, a general inverse problem is posed and a formulation to solve it is developed. It is shown that in many cases, the resulting equation is ill-conditioned and requires a special technique called regularization to make it amenable to stable numerical solution. Well-known techniques, such as Tikhonov regularization, and truncated singular value decomposition (TSVD) are discussed and demonstrated through examples. Commonly used gradient-based methods, such as the Newton's method, the conjugate gradient (CG) method, and the Levenberg-Marquardt algorithm are also discussed and demonstrated. Finally, the chapter delves into methods based on metaheuristics, namely simulated annealing and machine learning techniques based on neural networks and other approaches. Practical applications, such as optical tomography, are discussed to close the chapter.

Remaining Useful Life prediction of rolling bearings based on risk assessment and degradation state coefficient.

Abstract: Prediction of Remaining Useful Life (RUL) of bearings is very important for the condition-based maintenance of the rotating machinery. In order to predict the RUL more accurately, an approach of RUL prediction based on risk assessment and degradation state coefficient is proposed. The Mahalanobis Distance (MD) is calculated depending on the characteristics of vibration signals in the time domain and time-frequency domain. The features in the time-frequency domain are extracted by using Variational Mode Decomposition-Singular Value Decomposition (VMD-SVD). The monotonously increasing Health Indicator (HI) is obtained by using MD1-CUMSUM. The risk assessment is proposed to adaptively determine the thresholds of initial fault and failure, which is a trade-off between the false alarm rate and sensitivity. The RUL prediction of the testing bearing is completed based on the Genetic Algorithm-Support Vector Regression (GA-SVR) and Modified Health Indicator (MHI) with the degradation state coefficient. The proposed approach is verified by using two experimental vibration signal datasets. The results show that the proposed method has good capability to predict the RUL of rolling bearings.

A robust visual image encryption scheme based on controlled quantum walks

Abstract: Considering that traditional cryptographic systems are under threat of quantum computing, a visual image encryption algorithm based on controlled quantum is proposed. First, the plaintext image is encrypted by compressed sensing, and the compressed image is scrambled using Arnold to obtain the secret image. Then the segmented secret image is embedded into the carrier image by singular value decomposition (SVD) in accordance with different intensities to generate images with visual significance. In order to make the algorithm more secure, the controlled quantum walk is used to generate pseudorandom sequence to construct the measurement matrix of compressed sensing and the position of the secret image embedded into the carrier image. In addition, the use of SVD embedding can make the plaintext image flexible to choose the carrier image.

Double-Matrix Decomposition Image Steganography Scheme Based on Wavelet Transform with Multi-Region Coverage

Abstract: On the basis of ensuring the quality and concealment of steganographic images, this paper proposes a double-matrix decomposition image steganography scheme with multi-region coverage, to solve the problem of poor extraction ability of steganographic images under attack or interference. First of all, the cover image is transformed by multi-wavelet transform, and the hidden region covering multiple wavelet sub-bands is selected in the wavelet domain of the cover image to embed the secret information. After determining the hidden region, the hidden region is processed by Arnold transform, Hessenberg decomposition, and singular-value decomposition. Finally, the secret information is embedded into the cover image by embedding intensity factor. In order to ensure robustness, the hidden region selected in the wavelet domain is used as the input of Hessenberg matrix decomposition, and the robustness of the algorithm is further enhanced by Hessenberg matrix decomposition and singular-value decomposition. Experimental results show that the proposed method has excellent performance in concealment and quality of extracted secret images, and secret information is extracted from steganographic images attacked by various image processing attacks, which proves that the proposed method has good anti-attack ability under different attacks.