Dictionary Training Research Articles

Automatic identification of MS lesions based on local steering kernel features and sparse dictionary training

Automatic identification of MS lesions based on local steering kernel features and sparse dictionary training

Sparse representation for massive MIMO satellite channel based on joint dictionary learning

AbstractA constrained joint dictionary learning (CJDL) algorithm for high‐precision channel representation in massive multiple input multiple output (MIMO) satellite systems is proposed. Furthermore, taking into account the angular reciprocity of massive MIMO satellite systems, joint dictionary learning can establish a common support basis for both uplink and downlink. Previous deterministic dictionary has utilized deterministic basis, such as discrete Fourier transform (DFT) or orthogonal DFT (ODFT) basis, which tend to represent noise interference as part of channel characteristics. Furthermore, this deterministic dictionary is not able to adapt to dynamic communication environments. However, dictionary learning has shown the potential to significantly improve the accuracy of channel representation. Nevertheless, current research on training dictionary lacks analysis regarding constraints and boundary requirements, resulting in a suboptimal basis. To address this issue, conditional constraints associated with joint dictionary for channel representation are analysed. To screen for optimal basis, the joint dictionary is subject to constraints, including uplink and downlink constraints. Furthermore, the authors aim to quantify the maximum boundary of joint dictionary learning. Additionally, a joint dictionary updating method with singular value decomposition under constraint boundary conditions is proposed. Simulation results demonstrate that the proposed CJDL algorithm provides a more accurate and robust channel representation.

Vortex rope identification in Francis turbine based on cyclostationary extended dictionary learning

Francis turbine is one of the most commonly used hydraulic machinery for hydroelectric power generation, but its stability is frequently jeopardized by vortex rope. It is crucial to promptly and accurately detect vortex rope for the reliable operation of turbine unit. To address this challenge, the Cyclostationary Extended Dictionary Learning (CEDL) approach is proposed to identify vortex rope in Francis turbine. CEDL utilizes the Fisher discrimination criterion and low-rank constraints through the preprocessing and extending strategies to achieve accurate identification. Preprocessing specifically refers to the construction of filter for raw data based on cyclostationarity Then, the trained dictionary is used to extract features, and a linear classifier is developed based on these features. To evaluate the proposed method, vortex rope experiments are conducted in a Francis turbine under five conditions, and every condition includes four states: normal state, initial vortex rope, stable vortex rope and severe vortex rope. The results reveal three main findings. Firstly, the trained atoms in each sub-dictionary exhibit unique characteristics such as tonal frequency at 7 kHz and broadband frequency during 12–20 kHz, which are linked to the evolution stages of vortex rope. Secondly, the dictionary trained for a specific condition can achieve a classification accuracy of up to 90%, and even exceed 97% in some cases. Finally, the dictionary trained in a specific working condition can achieve relatively high classification accuracy in the scenarios of other working conditions.

Hausa in Hospitals: Exploring Healthcare Communications in Yariman Bakura Specialist Hospital Gusau, Zamfara State, Nigeria

This study investigates healthcare communication at Yariman Bakura Specialist Hospital (YBSH) Gusau in Zamfara state, focusing on the use of the Hausa language. A quantitative approach was employed through a self-administered survey questionnaire distributed to both patients and staff (n=200) at YBSH. The survey explored language preferences, dialect variations, challenges, and communication strategies. Purposive sampling ensured a representative sample of staff (doctors, nurses, etc.) and patients with diverse backgrounds. Data analysis focused on the prevalence of Hausa use, dialect variations, and staff perspectives on communication strategies and medical terminology. The study found Hausa as the dominant language (77% of staff, 64% of patients). However, dialect variations within Hausa (particularly Zamfarci) pose the challenges. The study identified a gap in using Hausa medical terminology. While staff primarily rely on interpreters (50%) and simplified explanations (45%), limited resources and knowledge restrict the use of medical terms in Hausa (5%). To improve communication, the study recommends developing a Hausa medical dictionary (focusing on Standard Hausa and Zamfarci dialects) and staff training programs on medical terminology. Additionally, a multi-pronged approach that includes interpreters, simplified explanations, and Hausa medical terminology is recommended for effective communication.

Target Image Processing Based on Super-resolution Reconstruction and Deep Machine Learning Algorithm

In dictionary-based single-frame image reconstruction algorithms, dictionaries rely on the design of artificial shallow features and are limited in their ability to represent image features. Therefore, this paper proposes a high-accuracy reconstruction method based on deep learning feature dictionary. This algorithm first uses a deep network to learn high-resolution and low-resolution training example images with deep features; Then co-train the feature dictionary under the super dense framework of the sparse dictionary; Finally, a single low-resolution image can be input and a super-resolution reconstruction can be performed using a dictionary. From the theoretical analysis, the introduction of deep network to extract the deep-level features of the image and its use in dictionary training is more beneficial to complement the high-frequency information in the low-resolution image. Experiments show that the proposed method achieves the best results in terms of both the peak signal-to-noise ratio and the gradient energy function of the reconstructed images. This shows that compared with traditional interpolation methods and some deep learning methods, the proposed method can recover image details to a high degree while preserving the original image damage information. This proves that the subjective visual and objective evaluation indicators of the algorithm presented in this article are higher than those of the comparative algorithm.

Exploiting sparsity of hyperspectral image: A novel approach for compressive hyperspectral image reconstruction using deep learning

Compressive hyperspectral imaging is an emerging technology that captures compressed two-dimensional (2D) measurements and subsequently reconstructs three-dimensional (3D) hyperspectral images, capitalizing on the sparsity prior. Numerous end-to-end networks have been proposed, yielding notable improvements in reconstruction speed and quality compared to traditional algorithms. However, these networks fall short in exploiting the sparsity of hyperspectral images in their direct mapping from 2D measurements to 3D hyperspectral images. To address this limitation, this paper introduces a novel approach for compressive hyperspectral image reconstruction using deep learning. The deep learning network comprises two components: the sparse coefficient recovery net and the dictionary training net. The dictionary training net incorporates a pre-trained dictionary, whose parameters can also be updated during the backpropagation process. Concurrently, the sparse coefficient recovery net generates sparse coefficients under this dictionary. Our approach draws inspiration from conventional dictionary learning algorithms, enabling effective utilization of signal sparsity. In comparison to several state-of-the-art methods, our method achieves superior reconstruction quality and demonstrates enhanced robustness against noise. The proposed approach is envisioned to offer fresh insights into hyperspectral images reconstruction using deep learning models.

Learning the sparse prior: Modern approaches

AbstractThe 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 Methods Statistical and Graphical Methods of Data Analysis > Modeling Methods and Algorithms Statistical Models > Nonlinear Models

Compressing of Medium- to Low-Rate Transform Residuals With Semi-Extreme Sparse Coding as an Alternate Transform in Video Coding

Modern hybrid video codecs split their video content into blocks. These blocks are predicted and the difference between original and predicted block is calculated. This residual block is transformed by a discrete cosine transform (DCT) from the pixel domain into the frequency domain and quantized by a dead-zone quantizer (DZQ), which removes high frequency signals, depending on the quantization parameter (QP). DZQ has an extended zone around zero, which acts as a noise-gate, removing noise as intended but also removing useful signals. DCT and DZQ are not the optimal solution, they create distortions and side effects at lower bitrates. In contrast, sparse coding does not have these problems at lower bitrates, but it compresses inefficient at higher bit rates. We propose a method to add sparse coding as an alternate transform, which is controlled by a rate-distortion optimization (RDO) decision, in a semi-extreme sparse coding (SESC) setup and uses a semi-extreme dictionary training (SEDT) process. It is integrated into the High Efficiency Video Coding (HEVC) test model HM-16.18 and screen content coding (HEVCSCC) test model HM-16.18+SCM-8.7. Experimental results demonstrate that the proposed method achieves a Bjontegaard rate difference (BD-rate) of up to 5.5% compared to the standard.

A fast dictionary-learning-based classification scheme using undercomplete dictionaries

In dictionary-learning-based classification methods, a given data point is classified based on its representation over one or possibly more learned dictionaries. The goal is to find dictionaries that minimize the classification error. Previous works aimed to train dictionaries with representation and classification powers by using overcomplete dictionaries and sparse coding. These approaches are computationally expensive and do not scale readily to problems with high dimensional data. This paper presents a dictionary-learning-based classification method with the primary goal of classification and not representation. We propose to train multiple undercomplete dictionaries (one for each class of the problem). Each dictionary approximates the given test data, and the one with the lowest reconstruction error determines the class. Singular value decomposition (SVD) is used to obtain a straightforward algorithm for the resulted optimization problem. Simulation results show that our method achieves a higher accuracy compared with a number of successful sparse representation based classification methods, while having a significantly lower computational cost.

Proposing a Continuous Spectral-Spatial Sparse Representation Model for Denoising Hyperspectral Images

Due to its high capability in acquiring spectral and spatial information, hyperspectral imaging technology has gained significant attention in remote sensing. However, in practice, it is impossible to avoid noise in hyper spectral images due to camera artifacts and the external environment during the acquisition and transmission process. The presence of noise in these images hinders the detection of subtle differences between different materials in the image. Therefore, it is crucial to minimize noise as much as possible before performing any analysis and interpretation. Removing noise from hyperspectral images is a crucial preprocessing step that enhances image quality for various applications, including object recognition and classification. The challenge arises when we need to remove additive white mean-spherohomogeneous Gaussian noise from the given image. Previous research has suggested that thinning the noise-free parts of the image can be effective in removing noise. This article aims to implement the method proposed in using a programming language. The method involves extracting intra-band structure and inter-band correlation while displaying the common tank and learning the common dictionary. In the continuous thin coding phase, the inter-band correlation is extracted to maintain the same structure and achieve spectrum continuity. In contrast, the intra-band structure is used to encode differences in the spatial structure of different bands. Furthermore, a joint dictionary training algorithm is used to obtain a dictionary that simultaneously describes the content of different bands. This ensures that the resulting dictionary preserves the inter-band correlations and enhances the noise-removal process.

English as a Foreign Language University (EFL) Students’ Perception of Dictionary Use and Training

This article examines students’ perceptions of dictionary use and training in English as a Foreign Language (EFL). The needs of English as a foreign language students in terms of specific dictionary types have received little attention. This research was conducted with 48 English as a foreign language students at a public university in Costa Rica over a year. The researcher used a survey (Appendix 1) to collect students’ opinions about the role of the dictionary in their major. Data indicates that: 1) not all dictionary types are known and used by students, 2) students favor the electronic dictionary over the paper dictionary, and 3) dictionary instruction is weak despite its perceived importance by learners. These conclusions are consistent with the literature surveyed. Furthermore, students’ opinions suggest that dictionary instruction and exposure have not been adequately approached in the language educational program. These results should advance new educational policies that promote the systematic and progressive inclusion of dictionary instruction across the study plan.

Self-adaptive seismic data reconstruction and denoising using dictionary learning based on morphological component analysis

Data reconstruction and data denoising are two critical preliminary steps in seismic data processing. Compressed Sensing states that a signal can be recovered by a series of solving algorithms if it is sparse in a transform domain, and has been well applied in the field of reconstruction, when, sparse representation of seismic data is the key point. Considering the complexity and diversity of seismic data, a single mathematical transformation will lead to incomplete sparse expression and bad restoration effects. Morphological Component Analysis (MCA) decomposes a signal into several components with outstanding morphological features to approximate the complex internal data structure. However, the representation ability of combined dictionaries is constrained by the number of dictionaries, and cannot be self-adaptively matched with the data features. Dictionary learning overcomes the limitation of fixed base function by training dictionaries that are fully suitable for processed data, but requires huge amount of time and considerable hardware cost. To solve the above problems, a new dictionary library (K-Singluar Value Decomposition learning dictionary and Discrete Cosine Transform dictionary) is hereby proposed based on the efficiency of fixed base dictionary and the high precision of learning dictionary. The self-adaptive sparse representation is achieved under the Morphological Component Analysis framework and is successfully applied to the reconstruction and denoising of seismic data. Real data tests have proved that the proposed method performs better than single mathematical transformation and other combined dictionaries.

Опыт проектирования электронного учебного словаря фразеологизмов делового английского языка

Importance. Presented the main results of creating an educational dictionary taking into account modern linguodidactic requirements to the process of designing reference aids. The main content of the research is related to the presentation of the process of creating an electronic training dictionary of phraseologisms related to business and economics. The novelty of the research lies in the rapid development of computer lexicography, which makes it necessary to use training dictionaries in teaching a foreign language for professional purposes. Materials and Methods. The study dictionaries of business English, textbooks on teaching English of business and economics sphere, as well as scientific literature and Internet sources are presented. Theoretical analysis of scientific literature on the research topic, the method of solid sampling, modeling and design methods, as well as elements of statistical analysis and pedagogical experiment are used as the main methods. Results and Discussion. During the experiment, the indicators improved more than twice, which suggests that the training dictionary helps students to memorize phraseological units better. Results obtained in the research seem to make a certain contribution to the solution of an important problem of formation of terminological competence of students of economic specialties, to the development of digital educational lexicography. Conclusion. The practical significance of the research lies in the possibility of using the developed lexicographic resource in the process of teaching students English in the sphere of business communication, as well as in the possibility of using the developed approaches to dictionary design in the practical activity of compiling electronic dictionaries of English for special (professional) purposes.

Joint Sparse Representation-based Single Image Super-Resolution for Remote Sensing Applications

Sparse representation-based single image super-resolution (SISR) methods use a coupled overcomplete dictionary trained from high-resolution (HR) images/ image patches. Since, remote sensing (RS) satellites capture images of large areas, these images usually have poor spatial resolution and obtaining an effective dictionary as such would be very challenging. Moreover, traditional patch-based sparse representation (PSR) models for reconstruction tend to give unstable sparse solution and produce visual artefact in the recovered images. To mitigate these problems, in this paper, we have proposed an adaptive joint sparse representation (JSR)-based SISR method that is dependent only on the input low-resolution (LR) image for dictionary training and sparse reconstruction. The new model combines patch-based local sparsity and group sparse representation (GSR)-based non-local sparsity in a single framework, which helps in stabilizing the sparse solution and improve the SISR results. Experimental results are evaluated both visually and quantitatively for several RGB and multispectral RS datasets, where the proposed method shows improvements in PSNR by 1–4 dB and 2–3 dB over the state-of-the-art sparse representation- and deep learning-based SR methods, respectively. Land cover classification applied on the super-resolved images further validate the advantages of the proposed method. Finally, for practical RS applications, we have performed parallel implementation in general purpose graphics processing units (GPGPU) and achieved significant speed-ups (30-40×) in the execution time.

Temperature and Emissivity Retrieval From Hyperspectral Thermal Infrared Data Using Dictionary-Based Sparse Representation for Emissivity

The separation of land surface temperature (LST) and land surface emissivity (LSE) is an ill-posed problem in thermal infrared (TIR) remote sensing. By building a new observation matrix to compress the LSE unknows and a dictionary training method to reconstruct complete LSE spectra, a new dictionary-based sparse representation for emissivity (DSRE) method has been proposed to retrieve LST and LSE from the atmospherically corrected hyperspectral TIR data. The proposed method fully utilizes the sparsity of compressed sensing and the empirical knowledge of the trained emissivity dictionary. The sensitivity analysis shows that the modeling accuracies of the proposed method are 0.215 <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">K</i> and 0.0060 for LST and LSE, respectively. Even with the instrument noise of 0.3 <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">K</i> and the uncertainties in atmospheric transmittance, atmospheric upwelling, and downwelling radiance of 10 %, the retrieval accuracies are 0.811 <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">K</i> for LST and 0.0241 for LSE, respectively. Then a field experiment was conducted to validate the proposed method, and a comparison was executed to three published methods, including ASTER temperature-emissivity separation (ASTERTES), linear spectral emissivity constraint TES (LSECTES), and iterative spectrally smooth TES (ISSTES). The accuracies of retrieved LST and spectral LSE are 1.41 <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">K</i> / 0.009, 2.57 <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">K</i> / 0.071, 1.59 <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">K</i> / 0.038, and 2.00 <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">K</i> / 0.077 for DSRE, ASTERTES, LSECTES, and ISSTES. In contrast to the three published methods, our proposed method is more accurate and effective than other published methods. Especially in the atmospheric absorption band, the proposed method has a strong anti-noise capability to the residuals of environmental downwelling radiance.

Theory of reduced biquaternion sparse representation and its applications

Traditional sparse representation models treat color image either represent color channels independently using the monochromatic model or concatenate color channels using the concatenation model. However, these two strategies cannot make full use of the correlation among the three color channels. In this paper, we propose a novel sparse representation model for color image based on reduced biquaternion. The advances of the proposed model are in two aspects: 1) compared with quaternion algebra, reduced biquaternion algebra is commutative algebra which has a low computation cost; 2) benefited from reduced biquaternion representation, the proposed model can treat the color image as a whole. We firstly extend the singular value decomposition (SVD) to the reduced biquaternion domain (RBSVD). The framework of the proposed model contains two major stages: first, reduced biquaternion orthogonal matching pursuit (RBOMP) is proposed in the stage of sparse coding; second, generalized K-means clustering for RBSVD (K-RBSVD) is proposed in the stage of dictionary training. Afterward, a new color image denoising algorithm is developed to demonstrate the effectiveness of the proposed model. The experimental results demonstrate that the proposed sparse representation model achieves state-of-the-art denoising performance in terms of both quantitative metrics and visual quality.

Representative‐discriminativedictionary learning algorithm for human action recognition using smartphone sensors

SUMMARYWith the advancement of mobile computing, understanding, and interpretation of human activities has become increasingly popular as an innovative human computer interaction application over the past few decades. This article presents a new scheme for action recognition based on sparse representation theory using a novel dictionary learning algorithm. This system employs two types of inertial signals from smartphones namely, accelerometer and gyroscope sensory data. Attainment of higher values of classification accuracy depends on the creation of effective dictionaries that completely retain the important features of every action while maintaining the least correlation with the features of other actions. Accordingly, in this research, we propose a new algorithm for learning dictionaries with two levels of dictionary training that aims at learning a compact, representative, and discriminative dictionary for each class. Unlike typical dictionary learning algorithms that aim at the creation of dictionaries that best represents the features of each class, our proposed algorithm incorporates a discriminative criterion that eventually produces better classification results. To validate the proposed framework, all the experiments were performed using three publicly available datasets.

Incoherent and Online Dictionary Learning Algorithm for Motion Prediction

Accurate model development and efficient representations of multivariate trajectories are crucial to understanding the behavioral patterns of pedestrian motion. Most of the existing algorithms use offline learning approaches to learn such motion behaviors. However, these approaches cannot take advantage of the streams of data that are available after training has concluded, and typically are not generalizable to data that they have not seen before. To solve this problem, this paper proposes two algorithms for learning incoherent dictionaries in an offline and online manner by extending the offline augmented semi-non-negative sparse coding (ASNSC) algorithm. We do this by adding a penalty into the objective function to promote dictionary incoherence. A trajectory-modeling application is studied, where we consider the learned atoms of the dictionary as local motion primitives. We use real-world datasets to show that the dictionaries trained by the proposed algorithms have enhanced representation ability and converge quickly as compared to ASNSC. Moreover, the trained dictionaries are well conditioned. In terms of pedestrian trajectory prediction, the proposed methods are shown to be on par (and often better) with the state-of-the-art algorithms in pedestrian trajectory prediction.

Local Visual and Global Deep Features Based Blind Stitched Panoramic Image Quality Evaluation Using Ensemble Learning

Image stitching technology aims to generate a Stitched Panoramic Image (SPI) by stitching multiple narrow-view images containing over-lapping regions. However, heterogeneous artifacts may be introduced during the process of image stitching, which affects the visual perception. To automatically and accurately evaluate the perceptual quality of SPIs, a novel local visual and global deep features based blind SPI quality evaluation method is proposed in this paper. To be specific, with the consideration that image stitching mainly destroys structure, texture and color information, we first design a color structure-texture joint dictionary trained from the constructed stitched-specific image patches dataset. Given an input SPI, its local visual and global deep features are extracted to characterize the stitched-specific distortions. For local visual features, the trained dictionary is employed to capture structure, texture and color distortions by sparse features extraction. Then, considering that sparse features are insensitive to weak structural distortions, weighted local binary pattern features are extracted to measure various weak distortions. For global perceptual features, deep features are extracted via a pre-trained convolutional neural network model to represent the high-level semantics. Finally, considering the diversity of the extracted features, an ensemble learning strategy is adopted to promote the generalization performance and prediction accuracy of the proposed model. Experimental results show that, compared with the conventional 2D and SPI quality measurement methods, the proposed method can measure the stitched-specific distortions more accurately, and is more coincident with subjective ratings.

Public food service communication field in the Chinese students’ linguistic consciousness: ethnocultural barriers and obstacles

The authors identify, classify and methodologically interpret communicative barriers and interferences arising for Chinese students in the sphere of catering, which is a fundamentally important culturally marked area of social and everyday communication. The relevance of this study is due to the significance and complexity of this social everyday sphere for the consciousness of foreign students, who face significant communicative barriers and obstacles that require methodological interpretation and consideration in the content of teaching Russian as a foreign language. The aims of this work are to identify and methodo- logically interpret communicative barriers of Chinese students in everyday communication (on the material of catering sphere), and to establish their correlations with the phenomena of lexical asymmetry. The main methods used in the research: mathematical processing of the obtained data, questionnaires, component analysis of vocabulary and comparison. The research material includes the results of the questionnaire aimed at identifying the barriers and obstacles encountered by Chinese students in the field of catering. It was determined that barriers and interference related to catering communication occurred for more than 75% of the students. Quantitative and meaningful processing of the questionnaire materials demonstrated that barriers and interference appeared in three main aspects: ethnographic, ethno- psychological, and ethnolinguistic. Methodologically oriented interpretation of communicative barriers is based on their correlation with the facts of the language. The integrative approach to the word taking into account its linguistic and extra-linguistic content determined which components of its structure translate ethnographic, ethnopsychological and ethnolinguistic differences of Russian and Chinese linguistic cultures that create communicative barriers. That is why the process of Russian language teaching considers conceptual, proper-lexical, semantic, connotative, background and contextual lacunas. Considering asymmetrical phenomena in the content of vocabulary teaching allows reducing the level of ethnographic, ethnopsychological and ethnolinguistic barriers and hindrances that arise for Chinese students in the sphere of catering. The prospects of the research include creation of the training dictionary “Gastronomic Culture Code in Language Vocabulary,” aimed at the Chinese students studying Russian.