Search

Video sequences exhibit significant nuisance variations (undesired effects) of speed of actions, temporal locations, and subjects' poses, leading to temporal-viewpoint misalignment when comparing two sets of frames or evaluating the similarity of two sequences. Thus, we propose Joint tEmporal and cAmera viewpoiNt alIgnmEnt (JEANIE) for sequence pairs. In particular, we focus on 3D skeleton sequences whose camera and subjects' poses can be easily manipulated in 3D. We evaluate JEANIE on skeletal Few-shot Action Recognition (FSAR), where matching well temporal blocks (temporal chunks that make up a sequence) of support-query sequence pairs (by factoring out nuisance variations) is essential due to limited samples of novel classes. Given a query sequence, we create its several views by simulating several camera locations. For a support sequence, we match it with view-simulated query sequences, as in the popular Dynamic Time Warping (DTW). Specifically, each support temporal block can be matched to the query temporal block with the same or adjacent (next) temporal index, and adjacent camera views to achieve joint local temporal-viewpoint warping. JEANIE selects the smallest distance among matching paths with different temporal-viewpoint warping patterns, an advantage over DTW which only performs temporal alignment. We also propose an unsupervised FSAR akin to clustering of sequences with JEANIE as a distance measure. JEANIE achieves state-of-the-art results on NTU-60, NTU-120, Kinetics-skeleton and UWA3D Multiview Activity II on supervised and unsupervised FSAR, and their meta-learning inspired fusion. [ABSTRACT FROM AUTHOR]

Different compression techniques necessitate the signal to be sparse or have a sparse representation in a suitable domain. Among these methods, compressed sensing is regarded as highly efficient. Sparse coding seeks to identify a sparse group of coefficients that effectively depicts a signal from a predetermined overcomplete dictionary. In the context of practical signal processing, quantization of sampled signals marks a crucial step in digitizing signals, and subsequently, realizing their high-efficiency transmission. In this study, we introduce a novel approach that integrates the challenges inherent in the approximation of sparse signals with coefficient quantization. In this study, a new method called Adaptive Quantized Iterative Thresholding is introduced, which utilizes Signal-to-Noise Ratio (SNR) to enhance classical iterative thresholding techniques for identifying sparse signal representations. Simulation outcomes demonstrate a notable enhancement in operational SNR performance when compared to Quantized Iterative Thresholding methods. [ABSTRACT FROM AUTHOR]

The kinematics of the human hand exhibit complex and diverse characteristics unique to each individual. Various techniques such as vision-based, ultrasonic-based, and data-glove-based approaches have been employed to analyze human hand movements. However, a critical challenge remains in efficiently analyzing and classifying hand grasp types based on time-series kinematic data. In this paper, we propose a novel sparse coding feature extraction technique based on dictionary learning to address this challenge. Our method enhances model accuracy, reduces training time, and minimizes overfitting risk. We benchmarked our approach against principal component analysis (PCA) and sparse coding based on a Gaussian random dictionary. Our results demonstrate a significant improvement in classification accuracy: achieving 81.78% with our method compared to 31.43% for PCA and 77.27% for the Gaussian random dictionary. Furthermore, our technique outperforms in terms of macro-average F1-score and average area under the curve (AUC) while also significantly reducing the number of features required. [ABSTRACT FROM AUTHOR]

Accuracy-optimized convolutional neural networks (CNNs) have emerged as highly effective models at predicting neural responses in brain areas along the primate ventral stream, but it is largely unknown whether they effectively model neurons in the complementary primate dorsal stream. We explored how well CNNs model the optic flow tuning properties of neurons in dorsal area MSTd and we compared our results with the Non-Negative Matrix Factorization (NNMF) model, which successfully models many tuning properties of MSTd neurons. To better understand the role of computational properties in the NNMF model that give rise to optic flow tuning that resembles that of MSTd neurons, we created additional CNN model variants that implement key NNMF constraints – non-negative weights and sparse coding of optic flow. While the CNNs and NNMF models both accurately estimate the observer's self-motion from purely translational or rotational optic flow, NNMF and the CNNs with nonnegative weights yield substantially less accurate estimates than the other CNNs when tested on more complex optic flow that combines observer translation and rotation. Despite its poor accuracy, NNMF gives rise to tuning properties that align more closely with those observed in primate MSTd than any of the accuracy-optimized CNNs. This work offers a step toward a deeper understanding of the computational properties and constraints that describe the optic flow tuning of primate area MSTd.

Image restoration (IR) attempts to recreate the original (ideal) scene from a degraded observation. The goal of image restoration is to avert the process of image deterioration that happens during image acquisition and processing. Blurring and noise are two common types of picture capture degradation. When the blurring function is unknown, the image restoration issue is referred to as blind image restoration. Existing methods of image restoration do not provide efficient results due to time complexity, computational complexity, large-scale scaling factor and limited input. Therefore, in order to overcome those problems, an Improved Graph Laplacian Regularization with Sparse Coding by integrating Internet of Things (IoT) is developed in this research. The process of removing de-noised portions of image to texture layer and cartoon layer is carried out by Morphological Component Analysis (MCA). An improved Graph Laplacian regularized method and Simultaneous Sparse Coding with Gaussian Scale Mixture (SSC-GSM) methods is performed on texture layer, cartoon layer and restored image. A Levin's dataset and a real-time dataset such as industrial manufacturing products are analyzed in this research. The proposed Graph Laplacian algorithm and sparse coding model are better efficient in identifying sharper texture information and getting a restored image. The proposed method proves that it has better performance in terms of Peak Signal-to-Noise Ratio (PSNR) of 35.82 and Structural Similarity Index Measure (SSIM) of 0.94 than the existing methods. [ABSTRACT FROM AUTHOR]

Existing discriminant analysis-synthesis dictionary pair learning (ASDPL) methods learn a structured analysis dictionary containing multiple class-specific analysis sub-dictionaries and a structured synthesis dictionary containing multiple class-specific synthesis sub-dictionaries. Although existing discriminant ASDPL methods achieve promising results in the field of pattern classification, most of them ignore the correlation between an analysis sub-dictionary and a synthesis sub-dictionary that belong to different classes, which may degrade the discriminative ability of their learned dictionary pairs. Moreover, most existing discriminant ASDPL methods do not give an explicit constraint to ensure that the reconstruction error of training samples under the joint action of a structured analysis dictionary and a structured synthesis dictionary is as small as possible, leading to insufficient representational ability of their learned dictionary pairs. To address these issues, we present a double-constrained structured discriminant analysis-synthesis dictionary pair learning (DCSDDPL) method. Specifically, we first design a class-specific analysis-synthesis sub-dictionary pair reconstruction constraint term to ensure that the reconstruction error of training samples of a class is as small as possible under the joint action of the analysis and synthesis sub-dictionaries belonging to the same class, which helps to improve the representational ability of the learned dictionary pair. Then, we design an analysis-synthesis sub-dictionary pair independence constraint term to eliminate the correlation between an analysis sub-dictionary and a synthesis sub-dictionary belonging to different classes so as to ensure the discriminative ability of the learned dictionary pair. Finally, we formulate the DCSDDPL model by integrating the two constraint terms into the basic discriminant analysis-synthesis dictionary pair learning model. Moreover, we design an optimization algorithm and use it to obtain the solution of the DCSDDPL model. We experimentally compare our method with five state-of-the-art dictionary learning methods, D-KSVD, LC-KSVD, FDDPL, DASDL and RA-DPL on the Extended Yale B, AR, PIE, CLD 22, Scene 15 and Caltech 101 datasets. The highest classification accuracy achieved by our method on these datasets is 99.13% and the highest F1-Score is 0.9906. The experimental results confirm that our method is effective for pattern classification. [ABSTRACT FROM AUTHOR]

Existing handcrafted keypoint detectors typically focus on designing specific local structures manually while ignoring whether they have enough flexibility to explore diverse visual patterns in an image. Despite the advancement of learning-based approaches in the past few years, most of them still rely on the availability of the outputs of handcrafted detectors as a part of training. In fact, such dependence limits their ability to discover various visual information. Recently, semi-handcrafted methods based on sparse coding have emerged as a promising paradigm to alleviate the above issue. However, the visual relationships between feature points have not been considered in the encoding stage, which may weaken the discriminative capability of feature representations for keypoint recognition. To tackle this problem, we propose a novel sparsity-guided discriminative feature representation (SDFR) method that attempts to explore the intrinsic correlations of keypoint candidates, thus ensuring the validity of characterizing distinctive and diverse structural information. Specifically, we first incorporate an affinity constraint into the feature representation objective, which jointly encodes all the patches in an image while highlighting the similarities and differences between them. Meanwhile, a smoother sparsity regularization with the Frobenius norm is leveraged to further preserve the similarity relationships of patch representations. Due to the differentiable property of this sparsity, SDFR is computationally feasible and effective for representing dense patches. Finally, we treat the SDFR model as multiple optimization sub-problems and introduce an iterative solver. During comprehensive evaluations on five challenging benchmarks, the proposed method achieves favorable performances compared with the state of the art in the literature. [ABSTRACT FROM AUTHOR]

High-resolution pathological images play a pivotal role in accurate disease diagnosis and are important in precision medicine. However, obtaining real-time high-resolution images faces challenges due to hardware limitations and scanning time constraints. Conventional image super-resolution restoration algorithms struggle to provide satisfactory results for pathological images, resulting in blurred and unrealistic restorations. In response to this challenge, this research proposes a pioneering approach by integrating Internet of Things (IoT) technology with Local Attention Dual Network (IOT-LAT) for super-resolution restoration of pathological images. The enhanced IOT-LATincorporates IoT-based data acquisition and processing, IoT-Enhanced non-local attention mechanisms, Gaussian constraint, and parameter-sharing strategies in up-sampling and down-sampling branches. This integration enables real-time super-resolution restoration of pathological images with improved accuracy. The reconstructed images exhibit a structural similarity of 0.914 and a peak signal-to-noise ratio of 30.84 dB. These results validate the effectiveness of the suggested approach in precisely reconstructing high-frequency details and improving modelling efficiency via a lightweight non-local attention mechanism enhanced by the Internet of Things. This work discusses IoT and non-local attention dual networks to improve pathological image super-resolution restoration. This allows for faster and more accurate disease diagnosis and treatment in precision medicine.

In this paper, we consider a sensor placement problem where sensors can move within a network over time. Sensor placement problem aims to select $K$ sensor positions from $N$ candidates where $K < N$. Most existing methods assume that sensor positions are static, i.e., they do not move, however, many mobile sensors like drones, robots, and vehicles can change their positions over time. Moreover, underlying measurement conditions could also be changed, which are difficult to cover with statically placed sensors. We tackle the problem by allowing the sensors to change their positions in their neighbors on the network. We dynamically determine the sensor positions based on graph signal sampling theory such that the non-observed signals on the network can be best recovered from the observations. For signal recovery, the dictionary is learned from a pool of observed signals. It is also used for the sensor position selection. In experiments, we validate the effectiveness of the proposed method via the mean squared error of the reconstructed signals. The proposed dynamic sensor placement outperforms the existing static ones for both synthetic and real data.

This paper introduces two variants of the backward replacement approach, T-BRe and H-BRe, aimed at enhancing the compression capabilities of forward greedy pursuit algorithms. The effectiveness of these methods is evaluated using three greedy pursuit algorithms (OMP, OOMP, and FBP) through simulations conducted on image and speech datasets with a mathematically generated dictionary. Additionally, a mathematical analysis of the H-BRe algorithm is provided to validate its effectiveness and address inherent limitations. Both analysis and simulations confirm that the H-BRe algorithm significantly improves the compression capabilities of forward greedy pursuit algorithms. The degree of enhancement varies depending on factors such as the forward modeling technique, stopping criterion, and the stochastic nature of the backward replacements. The high-performance (H) variants demonstrate superior performance across various modeling approaches, as evidenced by metrics such as PESQ and SSIM values, while the B and T variants also exhibit efficiency in specific scenarios. Despite these promising results, our methods encounter limitations such as time complexity, local error minimization, dependence on atom types, and the assumption of linear distribution. Future research directions include optimizing time complexity, balancing compression-quality trade-offs, globally minimizing replacement errors, exploring diverse atom types, and refining mathematical assumptions to further elucidate H-BRe’s performance characteristics.

Nocturnal enuresis is a bothersome condition that affects many children and their caregivers. Post-voiding systems is of little value in training a child into a correct voiding routing while existing pre-voiding systems suffer from several practical limitations, such as cumbersome hardware, assuming individual bladder shapes being universal, and being sensitive to sensor placement error. Methods: A low-voltage ultrasound system with machine learning has been developed in estimating bladder filling status. A custom-made flexible 1D transducer array has been excited by low-voltage coded pulses with a pulse compression technique for an enhanced signal-to-noise ratio. In order to minimize the negative influence of possible transducer misplacement, a multiple-position training strategy using machine learning has been adopted in this work. Three popular classification methods, KNN, SVM and sparse coding, have been utilized to classify the acquired different volumes ranging from 100 ml to 300 ml into two categories: low volume and high volume. The low-volume category requires no further action while the high-volume category triggers an alarm to alert the child and caregiver. Results: When the sensor placement is ideal, i.e., the position of the practical sensor placement is on spot with the trained position, the precision and recall of the classification using sparse coding are $0.957~\pm ~0.02$ and $0.958~\pm ~0.02$ , respectively. Even if the transducer array is misplaced by up to 4.5 mm away from the ideal location, the proposed system is able to maintain high classification accuracy (precision $\ge 0.75$ and recall $\ge 0.75$ ). Category: Early/Pre-Clinical Research Clinical and Translational Impact: The proposed ultrasound sensor system for nocturnal enuresis is of significant clinical and translational value as it addresses two major issues that limit the wide adoption of similar devices. Firstly, it offers enhanced safety as the entire system has been implemented in the lowvoltage domain. Secondly, the system features ample tolerance to sensor misplacement while maintaining high classification accuracy. These features combined provide a much more user-friendly environment for children and their caregivers than existing devices.

Ordinary independent component analysis cannot suitably handle the spherically symmetrical distributed subspaces that reflect the hue relations. Therefore, a method based on independent subspace analysis or multidimensional independent component analysis is presented for rotating a configuration in which color symbolism is visualized. Assuming two-dimensional spherically symmetric distributions for some pairs of dimensions, the proposed method aligns the axes into mutually independent spherically symmetrical spaces and sparse dimensions. The L2-norm of the data points in the spherically symmetric subspace is assumed to follow a Weibull distribution, and the distributions in the sparse dimensions are assumed to follow a generalized normal distribution. To process the heterogeneous dimensions/dimension pairs and to rotate an unwhitened configuration, the scaling parameters of dimensions/dimension pairs are estimated in addition to the orthonormal matrix by fully maximizing the likelihood function to rotate the space representing color symbolism. Two applications for this method are presented, whereby the configurations of color-emotion associations and color-shape associations obtained by correspondence analysis are rotated by the proposed method. The characteristics of these associations are clearly shown in the obtained configurations. The proposed method provides potentially advantageous options for rotating a configuration for the visualization of color symbolism.

Vision is the main sensory organ for human beings to contact and understand the objective world. The results of various statistical data show that more than 60% of all ways for human beings to obtain external information are through the visual system. Vision is of great significance for human beings to obtain all kinds of information needed for survival, which is the most important sense of human beings. The rapid growth of computer technology, image processing, pattern recognition, and other disciplines have been widely applied. Traditional image processing algorithms have some limitations when dealing with complex images. To solve these problems, some scholars have proposed various new methods. Most of these methods are based on statistical models or artificial neural networks. Although they meet the requirements of modern computer vision systems for feature extraction algorithms with high accuracy, high speed, and low complexity, these algorithms still have many shortcomings. For example, many researchers have used different methods for feature extraction and segmentation to get better segmentation results. Scale-invariant feature transform (SIFT) is a description used in the field of image processing. This description has scale invariance and can detect key points in the image. It is a local feature descriptor. A sparse coding algorithm is an unsupervised learning method, which is used to find a set of “super complete” basis vectors to represent sample data more efficiently. Therefore, combining SIFT and sparse coding, this article proposed an image feature extraction algorithm based on visual information to extract image features. The results showed that the feature extraction time of X algorithm for different targets was within 0.5 s when the other conditions were the same. The feature matching time was within 1 s, and the correct matching rate was more than 90%. The feature extraction time of Y algorithm for different targets was within 2 s. The feature matching time was within 3 s, and the correct matching rate was between 80 and 90%, indicating that the recognition effect of X algorithm was better than that of Y algorithm. It indicates the positive relationship between visual information and image feature extraction algorithm.

The nervous system is under tight energy constraints and must represent information efficiently. This is particularly relevant in the dorsal part of the medial superior temporal area (MSTd) in primates where neurons encode complex motion patterns to support a variety of behaviors. A sparse decomposition model based on a dimensionality reduction principle known as non-negative matrix factorization (NMF) was previously shown to account for a wide range of monkey MSTd visual response properties. This model resulted in sparse, parts-based representations that could be regarded as basis flow fields, a linear superposition of which accurately reconstructed the input stimuli. This model provided evidence that the seemingly complex response properties of MSTd may be a by-product of MSTd neurons performing dimensionality reduction on their input. However, an open question is how a neural circuit could carry out this function. In the current study, we propose a spiking neural network (SNN) model of MSTd based on evolved spike-timing-dependent plasticity and homeostatic synaptic scaling (STDP-H) learning rules. We demonstrate that the SNN model learns compressed and efficient representations of the input patterns similar to the patterns that emerge from NMF, resulting in MSTd-like receptive fields observed in monkeys. This SNN model suggests that STDP-H observed in the nervous system may be performing a similar function as NMF with sparsity constraints, which provides a test bed for mechanistic theories of how MSTd may efficiently encode complex patterns of visual motion to support robust self-motion perception.SIGNIFICANCE STATEMENT The brain may use dimensionality reduction and sparse coding to efficiently represent stimuli under metabolic constraints. Neurons in monkey area MSTd respond to complex optic flow patterns resulting from self-motion. We developed a spiking neural network model that showed MSTd-like response properties can emerge from evolving spike-timing-dependent plasticity with STDP-H parameters of the connections between then middle temporal area and MSTd. Simulated MSTd neurons formed a sparse, reduced population code capable of encoding perceptual variables important for self-motion perception. This model demonstrates that complex neuronal responses observed in MSTd may emerge from efficient coding and suggests that neurobiological plasticity, like STDP-H, may contribute to reducing the dimensions of input stimuli and allowing spiking neurons to learn sparse representations.

The sparse prior has been widely adopted to establish data models for numerous applications. In this context, most of them are based on one of three foundational paradigms: the conventional sparse representation, the convolutional sparse representation, and the multi‐layer convolutional sparse representation. When the data morphology has been adequately addressed, a sparse representation can be obtained by solving the sparse coding problem specified by the data model. This article presents a comprehensive overview of these three models and their corresponding sparse coding problems and demonstrates that they can be solved using convex and non‐convex optimization approaches. When the data morphology is not known or cannot be analyzed, it must be learned from training data, thereby formulating dictionary learning problems. This article addresses two different dictionary learning paradigms. In an unsupervised scenario, dictionary learning involves the alternating or joint resolution of sparse coding and dictionary updating. Another option is to create a recurrent neural network by unrolling algorithms designed to solve sparse coding problems. These networks can then be used in a supervised learning setting to facilitate the training of dictionaries via forward‐backward optimization. This article lists numerous applications in various domains and outlines several directions for future research related to the sparse prior. This article is categorized under:Statistical Learning and Exploratory Methods of the Data Sciences > Modeling MethodsStatistical and Graphical Methods of Data Analysis > Modeling Methods and AlgorithmsStatistical Models > Nonlinear Models [ABSTRACT FROM AUTHOR]

In conventional subspace clustering methods, affinity matrix learning and spectral clustering algorithms are widely used for clustering tasks. However, these steps face issues, including high time consumption and spatial complexity, making large-scale subspace clustering (LS 2 C) tasks challenging to execute effectively. To address these issues, we propose a large-scale subspace clustering method based on pure kernel tensor learning (PKTLS 2 C). Specifically, we design a pure kernel tensor learning (PKT) method to acquire as much data feature information as possible while ensuring model robustness. Next, we extract a small sample dataset from the original data and use PKT to learn its affinity matrix while simultaneously training a deep encoder. Finally, we apply the trained deep encoder to the original large-scale dataset to quickly obtain its projection sparse coding representation and perform clustering. Through extensive experiments on large-scale real datasets, we demonstrate that the PKTLS 2 C method outperforms existing LS 2 C methods in clustering performance. [ABSTRACT FROM AUTHOR]

To address the concerns with power consumption and processing efficiency in big-size data processing, sparse coding in computing-in-memory (CIM) architectures is gaining much more attention. Here, a novel Flash-based CIM architecture is proposed to implement large-scale sparse coding, wherein various matrix weight training algorithms are verified. Then, with further optimizations of mapping methods and initialization conditions, the variation-sensitive training (VST) algorithm is designed to enhance the processing efficiency and accuracy of the applications of image reconstructions. Based on the comprehensive characterizations observed when considering the impacts of array variations, the experiment demonstrated that the trained dictionary could successfully reconstruct the images in a 55 nm flash memory array based on the proposed architecture, irrespective of current variations. The results indicate the feasibility of using Flash-based CIM architectures to implement high-precision sparse coding in a wide range of applications. [ABSTRACT FROM AUTHOR]

Optical imaging of calcium signals in the brain has enabled researchers to observe the activity of hundreds-to-thousands of individual neurons simultaneously. Current methods predominantly use morphological information, typically focusing on expected shapes of cell bodies, to better identify neurons in the field-of-view. The explicit shape constraints limit the applicability of automated cell identification to other important imaging scales with more complex morphologies, e.g., dendritic or widefield imaging. Specifically, fluorescing components may be broken up, incompletely found, or merged in ways that do not accurately describe the underlying neural activity. Here we present Graph Filtered Temporal Dictionary (GraFT), a new approach that frames the problem of isolating independent fluorescing components as a dictionary learning problem. Specifically, we focus on the time-traces-the main quantity used in scientific discovery-and learn a time trace dictionary with the spatial maps acting as the presence coefficients encoding which pixels the time-traces are active in. Furthermore, we present a novel graph filtering model which redefines connectivity between pixels in terms of their shared temporal activity, rather than spatial proximity. This model greatly eases the ability of our method to handle data with complex non-local spatial structure. We demonstrate important properties of our method, such as robustness to morphology, simultaneously detecting different neuronal types, and implicitly inferring number of neurons, on both synthetic data and real data examples. Specifically, we demonstrate applications of our method to calcium imaging both at the dendritic, somatic, and widefield scales.

The kinematics of the human hand exhibit complex and diverse characteristics unique to each individual. Various techniques such as vision-based, ultrasonic-based, and data-glove-based approaches have been employed to analyze human hand movements. However, a critical challenge remains in efficiently analyzing and classifying hand grasp types based on time-series kinematic data. In this paper, we propose a novel sparse coding feature extraction technique based on dictionary learning to address this challenge. Our method enhances model accuracy, reduces training time, and minimizes overfitting risk. We benchmarked our approach against principal component analysis (PCA) and sparse coding based on a Gaussian random dictionary. Our results demonstrate a significant improvement in classification accuracy: achieving 81.78% with our method compared to 31.43% for PCA and 77.27% for the Gaussian random dictionary. Furthermore, our technique outperforms in terms of macro-average F1-score and average area under the curve (AUC) while also significantly reducing the number of features required.

(1) Background: The ability to recognize identities is an essential component of security. Electrocardiogram (ECG) signals have gained popularity for identity recognition because of their universal, unique, stable, and measurable characteristics. To ensure accurate identification of ECG signals, this paper proposes an approach which involves mixed feature sampling, sparse representation, and recognition. (2) Methods: This paper introduces a new method of identifying individuals through their ECG signals. This technique combines the extraction of fixed ECG features and specific frequency features to improve accuracy in ECG identity recognition. This approach uses the wavelet transform to extract frequency bands which contain personal information features from the ECG signals. These bands are reconstructed, and the single R-peak localization determines the ECG window. The signals are segmented and standardized based on the located windows. A sparse dictionary is created using the standardized ECG signals, and the KSVD (K-Orthogonal Matching Pursuit) algorithm is employed to project ECG target signals into a sparse vector–matrix representation. To extract the final representation of the target signals for identification, the sparse coefficient vectors in the signals are maximally pooled. For recognition, the co-dimensional bundle search method is used in this paper. (3) Results: This paper utilizes the publicly available European ST-T database for our study. Specifically, this paper selects ECG signals from 20, 50 and 70 subjects, each with 30 testing segments. The method proposed in this paper achieved recognition rates of 99.14%, 99.09%, and 99.05%, respectively. (4) Conclusion: The experiments indicate that the method proposed in this paper can accurately capture, represent and identify ECG signals. [ABSTRACT FROM AUTHOR]

One of the most well established brain principles, Hebbian learning, has led to the theoretical concept of neural assemblies. Based on it, many interesting brain theories have spawned. Palm's work implements this concept through multiple binary Willshaw associative memories, in a model that not only has a wide cognitive explanatory power but also makes neuroscientific predictions. Yet, Willshaw's associative memory can only achieve top capacity when the stored vectors are extremely sparse (number of active bits can grow logarithmically with the vector's length). This strict requirement makes it difficult to apply any model that uses this associative memory, like Palm's, to real data. Hence the fact that most works apply the memory to optimal randomly generated codes that do not represent any information. This issue creates the need for encoders that can take real data, and produce sparse representations - a problem which is also raised following Barlow's efficient coding principle. In this work, we propose a biologically-constrained network that encodes images into codes that are suitable for Willshaw's associative memory. The network is organized into groups of neurons that specialize on local receptive fields, and learn through a competitive scheme. After conducting auto- and hetero-association experiments on two visual data sets, we can conclude that our network not only beats sparse coding baselines, but also that it comes close to the performance achieved using optimal random codes. [ABSTRACT FROM AUTHOR]

This paper studies the convergence behaviour of dictionary learning via the Iterative Thresholding and K-residual Means (ITKrM) algorithm. On one hand, it is proved that ITKrM is a contraction under much more relaxed conditions than previously necessary. On the other hand, it is shown that there seem to exist stable fixed points that do not correspond to the generating dictionary, which can be characterised as very coherent. Based on an analysis of the residuals using these bad dictionaries, replacing coherent atoms with carefully designed replacement candidates is proposed. In experiments on synthetic data, this outperforms random or no replacement and always leads to full dictionary recovery. Finally, the question how to learn dictionaries without knowledge of the correct dictionary size and sparsity level is addressed. Decoupling the replacement strategy of coherent or unused atoms into pruning and adding, and slowly carefully increasing the sparsity level, leads to an adaptive version of ITKrM. In several experiments, this adaptive dictionary learning algorithm is shown to recover a generating dictionary from randomly initialized dictionaries of various sizes on synthetic data and to learn meaningful dictionaries on image data. [ABSTRACT FROM AUTHOR]

Level 3 automated driving mandates the installation of a driver monitoring system that monitors the driver's condition, such as drowsiness. Conventional drowsiness detection has been hampered by the mental and physical burden of wearing equipment and the time required for detection. In this study, we applied sparse coding to facial images captured with near‐infrared light to extract features of facial skin blood flow information related to drowsiness and to estimate drowsiness. As a result, skin blood flow features related to drowsiness were extracted in the nasal area and other areas, and drowsiness was estimated with 74.6% accuracy. © 2023 Institute of Electrical Engineer of Japan and Wiley Periodicals LLC. [ABSTRACT FROM AUTHOR]

With the spread of aerial laser bathymetry (ALB), seafloor topographies are being measured more frequently. Nevertheless, data deficiencies occur owing to seawater conditions and other factors. Conventional interpolation methods generally need to produce digital elevation models (DEMs) with sufficient accuracy. If the topographic features are considered as a basis, the DEM should be reproducible based on a combination of such features. The purpose of this study is to develop new DEM generation methods based on sparse modeling. Based on a review of the definitions of sparsity, we developed DEM generation methods based on a discrete cosine transform (DCT), DCT with elastic net, K-singular value decomposition (K-SVD), Fourier regularization, wavelet regularization, and total variation (TV) minimization, and conducted a comparative analysis. The developed methods were applied to artificially deficient DEM and ALB data, and their accuracy was evaluated. Thus, as a conclusion, we can confirm that the K-SVD method is appropriate when the percentage of deficiencies is low, and that the TV minimization method is appropriate when the percentage of deficiencies is high. Based on these results, we also developed a method integrating both methods and achieved an RMSE of 0.128 m. [ABSTRACT FROM AUTHOR]

The traditional image super-resolution (SR) reconstruction methods do not carry out sparse coding when performing SR reconstruction, resulting in poor structural similarity of image SR reconstruction. Therefore, this paper proposes a a super-resolution reconstruction method for polymorphic sparsity regularized images based on machine learning algorithms and builds a sparse representation model of polymorphic regularized images. For sparse representation models with different norm constraints, the orthogonal matching tracking in the greedy algorithm is used perform sparse coding to solve the sparse representation coefficient. The least square method in machine learning algorithm is used to solve the inverse operation of dictionary matrix update, construct the image super-resolution reconstruction minimization objective function, and solve the loss function between the super-resolution reconstruction image and the actual image, so as to complete the super-resolution reconstruction of polymorphic sparsity regularized image. The experimental results show that the edge recovery of the image under the proposed method is clearer, and the edge with more distinct edges and corners can be reconstructed. The Peak-Signal-to-Noise Ratio (PSNR) of the super-resolution reconstruction image is 60.5dB, and the Structural Similarity (SSIM) of the super-resolution reconstruction image is as high as 0.99, which effectively improves the image reconstruction effect. [ABSTRACT FROM AUTHOR]

To improve the detection for different fabric types and defect kinds, an approach based on K-Singular Value Decomposition (K-SVD) dictionary learning method is designed to detect fabric defects, which remains important and challenging in the field of textile engineering. The proposed method has two main parts, namely the training and detection. The 32 defect-free fabric samples were used to train the dictionary by K-SVD. The detection process mainly consists of image segmentation, reconstruction by the trained dictionary, and detection. To improve the detection speed, the patch size was applied for detecting the defects, and the learned dictionaries were trained offline. For selecting general patch size, three patch sizes of 16 $$ \times $$ 16 26 $$ \times $$ 26, and 36 $$ \times $$ 36 were used in the experiment. A comparative study found that the 26 $$ \times $$ 26 has a comprehensive performance. The size and location of the defects were denoted by the black rectangle. Experimental results on 55 fabric samples demonstrated that the proposed method can efficiently detect different kinds of fabric defects and fabric types based on false detection rate and the correct detection rate.

The mechanisms involved in transforming early visual signals to curvature representations in V4 are unknown. We propose a hierarchical model that reveals V1/V2 encodings that are essential components for this transformation to the reported curvature representations in V4. Then, by relaxing the often-imposed prior of a single Gaussian, V4 shape selectivity is learned in the last layer of the hierarchy from Macaque V4 responses. We found that V4 cells integrate multiple shape parts from the full spatial extent of their receptive fields with similar excitatory and inhibitory contributions. Our results uncover new details in existing data about shape selectivity in V4 neurons that with additional experiments can enhance our understanding of processing in this area. Accordingly, we propose designs for a stimulus set that allow removing shape parts without disturbing the curvature signal to isolate part contributions to V4 responses. [ABSTRACT FROM AUTHOR]

Compressed sensing theory is widely used to accurately reconstruct the original signal from a small number of random observations. Based on compressed sensing theory to achieve super-resolution reconstruction of images, the following two measures were taken on the basis of using the local sparsity prior of images. Firstly, by estimating the image degradation model, the K-singular value decomposition(K-SVD) algorithm was used to construct the over-complete dictionary pair, and based on the similarity of high and low resolution observations of the same image in the corresponding dictionary, the mapping relationship between high and low resolution images represented by the dictionary pair was brought into the objective function, avoiding the impact of downsampling and the difficulty of abstracting the fuzzy operator into matrix form on the solution. Secondly, an adaptive weighted adaptive weighting gradient projection for sparse reconstruction(AWGPSR) algorithm was proposed in the sparse coding of the super-resolution image to be sparse, which overcomes the traditional orthogonal matching pursuit(OMP) algorithm requires fixed sparsity in this step and can obtain more accurate sparse representation coefficients. The high-frequency component of the image can be reconstructed by combining the obtained sparse representation coefficients with the high-resolution dictionary, and the final image super-resolution reconstruction result can be obtained by fusing the reconstructed high-frequency component with the low-frequency part. The experimental results show that the proposed algorithm outperforms other related methods both in terms of subjective visual and objective evaluation indexes. [ABSTRACT FROM AUTHOR]

• A sparse representation method of heterogeneous information networks is proposed. • The heterogeneous information atoms are found by the sparse representation method. • The heterogeneous information atoms describe the main patterns of the heterogeneous information network. • The heterogeneous information atoms are helpful to understand the complex patterns of the heterogeneous information network. A complex network is a fundamental tool to describe real-world complex systems, with most real-world systems containing multiple object types and relationships that can be described as heterogeneous information networks. However, with the increasing network complexity, understanding the complex patterns and finding the meta paths or meta-structures of the heterogeneous information networks has become challenging. This paper proposes a sparse representation for heterogeneous information networks and extracts the heterogeneous information atoms that describe the basic connection pattern of the original heterogeneous information network. The heterogeneous information atoms help extract the main meta-paths or meta-structures and understand the complex patterns of the original heterogeneous information network. Furthermore, the heterogeneous information networks can be decomposed, dimension-reduced, and reconstructed through the heterogeneous information atoms. Extensive experimental results demonstrate that heterogeneous information atoms and sparse coding represent the basic connection pattern of real-world heterogeneous information networks. Indeed, the developed method can reconstruct a network with a recovery exceeding 90%. [ABSTRACT FROM AUTHOR]

Aiming at the curse of dimensionality and overfitting issues caused by multi-omics data for survival analysis, this paper proposed a survival analysis model based on multi-omics data and sparse variational autoencoder, called VAESCox. The algorithm combined the basic structure of variational autoencoder with sparse coding and survival analysis, trained the variational autoencoder to learn low-dimensional representations in the unsupervised stage, transfered the trained weights to the survival analysis model, fine-tunes and sparse encodes the passed weights in the supervised stage. Experimental results show that VAESCox model achieved higher C-index values in the ablation and comparison experiments on 8 different cancer types datasets. Compared with other four baseline survival analysis methods, the proposed algorithm not only mitigates the overfitting issue of multi-omics data integration, but also significantly improves survival prediction performance, indicating that the integration of different omics data is beneficial for accurate prediction of prognostic survival outcome. [ABSTRACT FROM AUTHOR]

A novel sparse encoding algorithm is developed to detect and study plasma instabilities automatically. This algorithm, called Elastic Random Mode Decomposition, is applied to the Mirnov coil signals of a dataset of 1291 discharges of the TJ-II stellarator, enabling the identification of the Alfvénic activity. In the presented approach, each signal is encoded as a collection of basic waveforms called atoms, drawn from a signal dictionary. Then the modes are identified using clustering and correlations with other plasma signals. The performance of the proposed algorithm is dramatically increased by using elastic net regularization and taking advantage of GPU architectures. Therefore the signal size and the number of dictionary elements are no longer limiting factors for encoding complex signals. Once the modes are retrieved from the shots, standard clustering and dimensionality reduction techniques are applied to obtain a 2D map featuring of the physical mode characteristics of this subset of TJ-II shots. The clustering features consider the relationship with the plasma current $I_\mathrm p$ , the diamagnetic energy W , and inverse square root of electron density $1/\sqrt{n}$ , profiling different subtypes of Alfvénic activity. The proposed algorithm can potentially create large databases of labeled modes with unprecedented detail.

Remote blood pressure measurement using visible images facilitates routine blood pressure monitoring and leads to early detection of hypertension, a risk factor for lifestyle diseases. The previous study that attempted to estimate blood pressure by applying CNN to facial thermal images found that facial images contain two types of features, physiological responses and expression changes, which need to be separated. In contrast, we found that these features could be separated by using sparse coding on facial thermal images. This study used sparse coding to extract physiological response areas during acute blood pressure fluctuations from facial visible images by examining preprocessing. The results indicated that sparse coding and the proposed preprocessing methods for images were effective. © 2023 Institute of Electrical Engineer of Japan and Wiley Periodicals LLC. [ABSTRACT FROM AUTHOR]

Considering practical device-free localization (DFL), the localization precision is usually proportional to the sensor density, with an ordinary arrangement. A better localization performance generally indicates a higher density of the sensor nodes that need to be deployed. To overcome this problem, we propose a new framework for dispatching sensor nodes. Regarding this framework, only few sensor nodes are applied to achieve an excellent localization performance. We consider the DFL problem as a sparse coding problem. To maintain the convexity of the cost function for more accurate solutions, we introduce the generalized minimax-concave (GMC) regularization to approximate the ℓ 0 -norm regularization. The global optimal solution can be identified by adopting the forward-backward splitting algorithm (FBS). Furthermore, the localization error is further decreased by the proposed adaptive relaxation localization (ARL) criteria for target localization. We tackled two experimental scenes in a real laboratory and compared the performance of the proposed algorithm with that of other algorithms using different regularizations. The experimental results show 100 % grid localization accuracy under the 0.5 × 0.5 m grid scene. After adopting the ARL criteria, the average localization error decreased from 0.098 to 0.053 m in the 0.25 × 0.25 m grid scene, with an increased rate of 45.9 % . This is the best performance compared to state-of-the-art framework. [ABSTRACT FROM AUTHOR]

To improve the detection for different fabric types and defect kinds, an approach based on K-Singular Value Decomposition (K-SVD) dictionary learning method is designed to detect fabric defects, which remains important and challenging in the field of textile engineering. The proposed method has two main parts, namely the training and detection. The 32 defect-free fabric samples were used to train the dictionary by K-SVD. The detection process mainly consists of image segmentation, reconstruction by the trained dictionary, and detection. To improve the detection speed, the patch size was applied for detecting the defects, and the learned dictionaries were trained offline. For selecting general patch size, three patch sizes of 16 × 16 26 × 26, and 36 × 36 were used in the experiment. A comparative study found that the 26 × 26 has a comprehensive performance. The size and location of the defects were denoted by the black rectangle. Experimental results on 55 fabric samples demonstrated that the proposed method can efficiently detect different kinds of fabric defects and fabric types based on false detection rate and the correct detection rate. [ABSTRACT FROM AUTHOR]

Neural networks that are based on the unfolding of iterative solvers as LISTA (Learned Iterative Soft Shrinkage), are widely used due to their accelerated performance. These networks, trained with a fixed dictionary, are inapplicable in varying model scenarios, as opposed to their flexible non-learned counterparts. We introduce, Ada-LISTA, an adaptive learned solver which receives as input both the signal and its corresponding dictionary, and learns a universal architecture to serve them all. This scheme allows solving sparse coding in linear rate, under varying models, including permutations and perturbations of the dictionary. We provide an extensive theoretical and numerical study, demonstrating the adaptation capabilities of our approach, and its application to the task of natural image inpainting. [ABSTRACT FROM AUTHOR]

Energy disaggregation, or Non-Intrusive Load Monitoring (NILM), is a technique that predicts the consumption levels of individual appliances from only the main signal in the building. Various methods have been proposed to solve this problem, including sparse coding (SC), which offers great advantages due to its ability to capture complex patterns in data. However, a challenging aspect of NILM is that data containing appliance-level information is scarce. Moreover, the houses that the models are tested on might be from a different population than the training data, thus resulting in a domain shift. Therefore, we need to develop approaches that are adapted to training data scarcity through the use of transfer learning (TL). In this paper, we test several TL approaches on SC models with the aim of discriminative energy disaggregation (DD). We propose four TL SC models: "Transfer SC+DD", "SC source, DD target", "Transfer Deep SC" and "Transfer Variational SC". We compare these models with 4 models that do not apply domain adaptation. We show that TL-based models achieve similar and sometimes better results than the regular models. We also show that further training on only the aggregate signal of some target domain houses improves the performance of a SC model trained only on the source. [ABSTRACT FROM AUTHOR]

To address the concerns with power consumption and processing efficiency in big-size data processing, sparse coding in computing-in-memory (CIM) architectures is gaining much more attention. Here, a novel Flash-based CIM architecture is proposed to implement large-scale sparse coding, wherein various matrix weight training algorithms are verified. Then, with further optimizations of mapping methods and initialization conditions, the variation-sensitive training (VST) algorithm is designed to enhance the processing efficiency and accuracy of the applications of image reconstructions. Based on the comprehensive characterizations observed when considering the impacts of array variations, the experiment demonstrated that the trained dictionary could successfully reconstruct the images in a 55 nm flash memory array based on the proposed architecture, irrespective of current variations. The results indicate the feasibility of using Flash-based CIM architectures to implement high-precision sparse coding in a wide range of applications.

(1) Background: The ability to recognize identities is an essential component of security. Electrocardiogram (ECG) signals have gained popularity for identity recognition because of their universal, unique, stable, and measurable characteristics. To ensure accurate identification of ECG signals, this paper proposes an approach which involves mixed feature sampling, sparse representation, and recognition. (2) Methods: This paper introduces a new method of identifying individuals through their ECG signals. This technique combines the extraction of fixed ECG features and specific frequency features to improve accuracy in ECG identity recognition. This approach uses the wavelet transform to extract frequency bands which contain personal information features from the ECG signals. These bands are reconstructed, and the single R-peak localization determines the ECG window. The signals are segmented and standardized based on the located windows. A sparse dictionary is created using the standardized ECG signals, and the KSVD (K-Orthogonal Matching Pursuit) algorithm is employed to project ECG target signals into a sparse vector–matrix representation. To extract the final representation of the target signals for identification, the sparse coefficient vectors in the signals are maximally pooled. For recognition, the co-dimensional bundle search method is used in this paper. (3) Results: This paper utilizes the publicly available European ST-T database for our study. Specifically, this paper selects ECG signals from 20, 50 and 70 subjects, each with 30 testing segments. The method proposed in this paper achieved recognition rates of 99.14%, 99.09%, and 99.05%, respectively. (4) Conclusion: The experiments indicate that the method proposed in this paper can accurately capture, represent and identify ECG signals.