Regularization Algorithm Research Articles

3D Path Planning of the Solar Powered UAV in the Urban-Mountainous Environment with Multi-Objective and Multi-Constraint Based on the Enhanced Sparrow Search Algorithm Incorporating the Levy Flight Strategy

In response to practical application challenges in utilizing solar-powered unmanned aerial vehicle (UAV) for remote sensing, this study presents a three-dimensional path planning method tailored for urban-mountainous environment. Taking into account constraints related to the solar-powered UAV, terrain, and mission objectives, a multi-objective trajectory optimization model is transferred into a single-objective optimization problem with weight factors and multi-constraint and is developed with a focus on three key indicators: minimizing trajectory length, maximizing energy flow efficiency, and minimizing regional risk levels. Additionally, an enhanced sparrow search algorithm incorporating the Levy flight strategy (SSA-Levy) is introduced to address trajectory planning challenges in such complex environments. Through simulation, the proposed algorithm is compared with particle swarm optimization (PSO) and the regular sparrow search algorithm (SSA) across 17 standard test functions and a simplified simulation of urban-mountainous environments. The results of the simulation demonstrate the superior effectiveness of the designed improved SSA based on the Levy flight strategy for solving the established single-objective trajectory optimization model.

Investigation of spatial resolution of electrical capacitance tomography based on the electromagnetic momentum (ECT-EMM)

Electrical capacitance tomography (ECT) is a permittivity imaging method widely used in industrial inspection. The equations described by the ECT technique are nonlinear and ill-posed, which results in low image resolution. ECT can be considered an imaging method based on the Green’s reciprocity theorem, an energetic reciprocity theorem. ECT detects scalars, i.e. capacitances. Electromagnetic fields have both ‘energy’ and ‘momentum.’ In recent years, the electromagnetic momentum reciprocity theorem has enriched the electromagnetic reciprocity theorem. The electromagnetic momentum reciprocity theorem is an imaging method that detects vectors, i.e. capacitance gradients. Vectors contain richer information than scalars; thus, electrical capacitance tomography based on electromagnetic momentum (ECT-EMM) methods is expected to improve the resolution of permittivity imaging. This paper briefly describes the principle of the ECT-EMM technique for image reconstruction using sensitivity matrix gradient and capacitance gradient. Tikhonov regularisation algorithm is applied. The two methods, with and without capacitance measurements, are used to evaluate imaging resolution. Under different numbers of pixels and electrodes, typical permittivity distributions are used for reconstruction, and correlation coefficients are calculated. Simulations and experiments show that the ECT-EMM technique recognises object boundaries more clearly with high noise immunity. Five quality measures are used to evaluate the performance of the point spread function without capacitance measurements. Compared to ECT, the ECT-EMM technique is more sensitive to the central region away from the electrodes, recognises smaller minimum objects, and has smaller shape deformation.

A Second-Order Generalized Total Variation with Improved Alternating Direction Method of Multipliers Algorithm for Electrical Impedance Tomography Reconstruction

The issue of Electrical Impedance Tomography (EIT) is a well-known inverse problem that presents challenging characteristics. In order to address the difficulties associated with ill-conditioned inverses, regularization methods are typically employed. One commonly used approach is total variation (TV) regularization, which has shown effectiveness in EIT. In order to meet the requirements of real-time tracking, it is essential to acquire fast and reliable algorithms for image reconstruction. Therefore, we present a modified second-order generalized regularization algorithm that enables more-accurate reconstruction of organ boundaries and internal structures, to reduce EIT artifacts, and to overcome the inability of the conventional Tikhonov regularization method in solving the step effect of the medium boundary. The proposed algorithm uses the improved alternating direction method of multipliers (ADMM) to tackle this optimization issue and adopts the second-order generalized total variation (SOGTV) function with strong boundary-preserving features as the regularization generalization function. The experiments are based on simulation data and the physical model of the circular water tank that we developed. The results showed that SOGTV regularization can improve image realism compared with some classic regularization.

Dual-wavelength Fourier ptychographic microscopy for topographic measurement.

Topographic measurements of micro- or nanostructures are essential in cutting-edge scientific disciplines such as optical communications, metrology, and structural biology. Despite the advances in surface metrology, measuring micron-scale steps with wide field of view (FOV) and high-resolution remains difficult. This study demonstrates a dual-wavelength Fourier ptychographic microscopy for high-resolution topographic measurement across a wide FOV using an aperture scanning structure. This structure enables the capture of a three-dimensional (3D) sample's scattered field with two different wavelength lasers, thus allowing the axial measurement range growing from nano- to micro-scale with enhanced lateral resolution. To suppress the unavoidable noises and artifacts caused by temporal coherence, system vibration, etc., a total variation (TV) regularization algorithm is introduced for phase retrieval. A blazed grating with micron-scale steps is used as the sample to validate the performance of our method. The agreement between the high-resolution reconstructed topography with our method and that with atomic force microscopy verified the effectiveness. Meanwhile, numerical simulations suggest that the method has the potential to characterize samples with high aspect-ratio steps.

Modeling of a high gain two stage pHEMT LNA using ANN with Bayesian regularization algorithm

Modeling of a high gain two stage pHEMT LNA using ANN with Bayesian regularization algorithm

A Forward−Backward Splitting Equivalent Source Method Based on S−Difference

The regularization method has a direct impact on the accuracy of the reconstructed sound field in the process of inverse calculation of near−field acoustic holography using the equivalent source method. To expand the frequency range of sound field reconstruction and improve computational accuracy, a forward-backward splitting equivalent source method based on s−difference was proposed, which uses the ratio of the output results of the broadband acoustic holography algorithm as the regularization parameter. Numerical simulations of single source and coherent source sound fields were conducted under different frequency conditions to analyze the performance of the forward-backward splitting regularization algorithm based on s−difference, and experimental verification was performed. The simulation results show that the proposed method can accurately reconstruct the sound field in a wider frequency range, and has high accuracy in reconstructing the sound field of low−frequency coherent sources. The experimental results demonstrate the accuracy and effectiveness of this method in reconstructing mid−to−low−frequency sound fields.

Hardware and Software Design and Implementation of Surface-EMG-Based Gesture Recognition and Control System

The continuous advancement of electronic technology has led to the gradual integration of automated intelligent devices into various aspects of human life. Motion gesture-based human–computer interaction systems offer abundant information, user-friendly functionalities, and visual cues. Surface electromyography (sEMG) signals enable the decoding of muscle movements, facilitating the realization of corresponding control functions. Considering the inherent instability and minuscule nature of sEMG signals, this thesis proposes the integration of a dynamic time regularization algorithm to enhance gesture recognition detection accuracy and real-time system performance. The application of the dynamic time warping algorithm allows the fusion of three sEMG signals, enabling for the calculation of similarity between the sample and the model. This process facilitates gesture recognition and ensures effective communication between individuals and the 3D printed prosthesis. Utilizing this algorithm, the best feature model was generated by amalgamating six types of gesture classification model. A total of 600 training and evaluation experiments were performed, with each movement recognized 100 times. The experimental tests demonstrate that the accuracy of gesture recognition and prosthetic limb control using the temporal dynamic regularization algorithm achieves an impressive 93.75%, surpassing the performance of the traditional threshold control switch.

Online regularized learning algorithm for functional data

In recent years, functional linear models have attracted growing attention in statistics and machine learning for recovering the slope function or its functional predictor. This paper considers online regularized learning algorithm for functional linear models in a reproducing kernel Hilbert space. It provides convergence analysis of excess prediction error and estimation error with polynomially decaying step-size and constant step-size, respectively. Fast convergence rates can be derived via a capacity dependent analysis. Introducing an explicit regularization term extends the saturation boundary of unregularized online learning algorithms with polynomially decaying step-size and achieves fast convergence rates of estimation error without capacity assumption. In contrast, the latter remains an open problem for the unregularized online learning algorithm with decaying step-size. This paper also demonstrates competitive convergence rates of both prediction error and estimation error with constant step-size compared to existing literature.

Research on Piano Curriculum Education and Its Performance Ecosystem Based on Network Flow Optimization

Abstract This paper investigates music education, where an efficient and accurate performance evaluation system in the piano teaching and performance ecosystem is increasingly becoming an essential tool for improving teaching quality and performance level. The objective evaluation of students’ performance skills can be achieved by carefully analyzing piano performances using the network flow optimization technique. This technique optimizes the performance evaluation system’s audio recognition ability by analyzing the piano audio signal and solving the multi-constraint nonlinear optimization problem in a limited time domain. This paper establishes a network flow optimization model, applies the multi-constraint nonlinear optimization technique, and combines the non-negative matrix decomposition and dynamic time regularization algorithm to analyze the piano performance for experiments. After optimization processing, hundreds of piano audio samples were collected, and the audio recognition accuracy was improved by 20%. By optimizing and processing the audio signals from the network stream, the evaluation system could detect polyphony more accurately and track the musical score effectively, improving accuracy and efficiency. Using the non-negative matrix decomposition algorithm, the accuracy of detecting polyphony can reach 85%, while the dynamic temporal regularization algorithm can match the position of the musical score with 95% accuracy. The accuracy of piano performance evaluation is optimized by this network flow optimization method, providing new technical means for music education, and promoting the quality of teaching and performance.

Dynamic Correlation Learning and Regularization for Multi-Label Confidence Calibration.

Modern visual recognition models often display overconfidence due to their reliance on complex deep neural networks and one-hot target supervision, resulting in unreliable confidence scores that necessitate calibration. While current confidence calibration techniques primarily address single-label scenarios, there is a lack of focus on more practical and generalizable multi-label contexts. This paper introduces the Multi-Label Confidence Calibration (MLCC) task, aiming to provide well-calibrated confidence scores in multi-label scenarios. Unlike single-label images, multi-label images contain multiple objects, leading to semantic confusion and further unreliability in confidence scores. Existing single-label calibration methods, based on label smoothing, fail to account for category correlations, which are crucial for addressing semantic confusion, thereby yielding sub-optimal performance. To overcome these limitations, we propose the Dynamic Correlation Learning and Regularization (DCLR) algorithm, which leverages multi-grained semantic correlations to better model semantic confusion for adaptive regularization. DCLR learns dynamic instance-level and prototype-level similarities specific to each category, using these to measure semantic correlations across different categories. With this understanding, we construct adaptive label vectors that assign higher values to categories with strong correlations, thereby facilitating more effective regularization. We establish an evaluation benchmark, re-implementing several advanced confidence calibration algorithms and applying them to leading multi-label recognition (MLR) models for fair comparison. Through extensive experiments, we demonstrate the superior performance of DCLR over existing methods in providing reliable confidence scores in multi-label scenarios.

Robust Sparse Recovery Based Vehicles Location Estimation in Intelligent Transportation System

An architecture for vehicle position estimation is introduced in this paper to tackle the issue of vehicles positioning in traffic congestion of intelligent transportation system (ITS). The introduced architecture is related to three unmanned aerial vehicles (UAVs) equipped with uniform linear array (ULA), ITS center and the terminal of position estimation, in which the UAV collect data from the ULA, the ITS center store data and the terminal is responsible for executing the corresponding direction of arrival (DOA) estimation algorithm to estimate the vehicle position. In the introduced architecture, DOA estimation is the crucial issue, hence a robust sparse recovery framework based on the optimal weighted subspace fitting is put forward for DOA estimation in the presence of direction-dependent unknown mutual coupling. Firstly, a data model can be obtained by a new transformation approach. Then, a sparse recovery algorithm based on the weighted subspace fitting is used to estimate the desired DOAs. The proposed DOA estimation algorithm can achieve superior performance in both coherent and incoherent signals in the environment of direction-dependent mutual coupling, and this environment often occurs in traffic congestion. Based on the DOA estimation results obtained from the proposed efficient regularized sparse recovery algorithm, the position of vehicles is final estimated by a weighted three-points cross-positioning method. The results of various simulation experiments fully demonstrate the robustness and superiority of the proposed architecture and algorithm.

ANN for the prediction of isobutylene dimerization through catalytic distillation for a preliminary energy and environmental evaluation

<abstract> <p>This study aimed to develop an artificial neural network (ANN) capable of predicting the molar concentration of diisobutylene (DIB), 3, 4, 4-trimethyl-1-pentene (DIM), and tert-butyl alcohol (TBA) in the distillate and residue streams within three specific columns: reactive (CDC), high pressure (ADC), and low pressure (TDC). The process simulation was conducted using DWSIM, an open-source platform. Following its validation, a sensitivity analysis was performed to identify the operational variables that influenced the molar fraction of DIB, DIM, and TBA in the outputs of the three columns. The input variables included the molar fraction of isobutylene (IB) and 2-butene (2-Bu) in the butane (C4) feed, the temperature of the C4 and TBA feeds, and the operating pressure of the CDC, ADC, and TDC columns. The network's design, training, validation, and testing were performed in MATLAB using the Neural FittinG app. The network structure was based on the Bayesian regularization (BR) algorithm, that consisted of 7 inputs and seven outputs with 30 neurons in the hidden layer. The designed, trained, and validated ANN demonstrated a high performance, with a mean squared error (MSE) of 0.0008 and a linear regression coefficient (R) of 0.9946. The statistical validation using an analysis of variance (ANOVA) (<italic>p</italic>-value > 0.05) supported the ANN's capability to reliably predict molar fractions. Future research will focus on the in-situ validation of the predictions and explore hybrid technologies for energy and environmental optimization in the process.</p> </abstract>

Research on deep learning-based action recognition and quantitative assessment method for sports skills

Abstract The current sports training lacks data-based scientific training tools, and the use of action recognition technology to collect and mine sports data can effectively identify and evaluate sports skill actions. In this paper, a Transformer-based convolutional neural human action recognition network is proposed, which integrates the C3D convolutional network with the visual Transformer structure, using the 3D convolutional kernel for the extraction of time-domain features and using the Transformer network to accurately classify the feature sequences. The OpenPose algorithm is used to extract the essential points of the skeletal joints to estimate the human action posture. Through the dynamic time regularization algorithm, athletes’ sports movements are matched with standard movements to achieve a quantitative assessment of sports skill movements. The experimental results show that the method in this paper has better performance than similar neural network models in the task of sports action recognition and evaluation, and its class average accuracy mAP value and GFLOPs/V value are 0.9291 and 25.01, respectively, which substantially improves the recognition efficiency of sports skill actions.

Simultaneous determination of 2-D distributions of temperature and absorption coefficient in large-scale pulverized-coal-fired boilers by flame images processing

This paper proposes a new algorithm that can simultaneously measure the two-dimensional temperature distribution and absorption coefficient distribution of a pulverized-coal-fired (PCF) boiler furnace. On the base of optimal estimation of the uniform absorption and scattering coefficients as well as 2-D temperature distribution from the color radiation images of the flame by the Tikhonov regularization algorithm, the product of the absorption coefficient and the absolute blackbody emission intensity forms a new unknown distribution of emission source, which is also reconstructed from the color radiation images of the flame by the Tikhonov method. Then the distribution of absorption coefficient can be obtained from the new distribution of emission source as the 2-D temperature distribution has been known. Experimental investigation has been conducted on a 660 MWe four corner, tangentially firing, PCF boiler. In the flame detectors, a three-band pass filter is added in the CCD industrial camera to improve the monochromaticity of the original flame image. The temperature and absorption coefficient distributions of the cross-sections at three detection layers under a load of 330 MW and at the highest layer under a load of 500 MW are obtained. In order to verify the reliability of the algorithm, the flame radiation intensity images calculated from the 2-D distributions of temperature and absorption coefficient reconstructed at the highest layer under the load of 330 MW are taken as measurement data, the algorithm is repeated and the results show that the 2-D temperature distribution reconstructed is similar to the input one, with relative error less than 1% between their mean temperatures, and relative error less than 10% between the reconstructed and input mean values of absorption and scattering coefficients. It demonstrates that the algorithm and technology proposed in this paper have important application prospects in the monitoring, diagnosis and control of PCF combustion processes.

Adaptive Bilateral-Total-Variation Regularization Algorithm for Enhancing HY2-SCAT Data

Adaptive Bilateral-Total-Variation Regularization Algorithm for Enhancing HY2-SCAT Data

Hybrid machine learning approach for accurate and expeditious 3D scanning to enhance rapid prototyping reliability in orthotics using RSM-RSMOGA-MOGANN

Abstract This study aims to develop a multidisciplinary artificial hybrid machine learning (AHML) approach to reduce the scanning time (ST) of the human wrist and improve the accuracy of 3D scanning for anthropometric data collection. A systematic AHML approach was deployed to scan the human wrist distal end optimally using a portable SENSE 2.0 3D scanner. A central composite design (CCD) matrix was developed for three input variables; light intensity (LI = 12–20 W/m2), capture angle (CA = 10°–50°), and scanning distance (SD = 10–20 inches) for executing the experimental runs. For accuracy evaluation, the wrist perimeter on the distal end was checked using CREO Parametric software for wrist perimeter error (WPE). Various AHML tools were developed using: response surface methodology (RSM), multi-objective genetic algorithm RSM, and multi-objective genetic algorithm neural networking (MOGANN). The optimal process parameters recommended by the hybrid tools were experimentally validated for their prediction accuracy. The MOGANN approach combined with the Bayesian regularization algorithm (trainabr) provided the best mutual combination of optimal ST = 20.072 sec and WPE = 0.375 cm corresponding to LI = 12.001 W/m2, CA = 29.428°, and SD = 18.214 inch, with a significant percentage reduction of 55.83% in WPE. Executing 3D scanning of the human wrist over the optimized process parameters predicted by AHML tools will ensure the availability of precise scans for the rapid prototyping of customized orthotic devices in a reliable manner.

Three-parameter Regularization Algorithm for Pseudo-solution of Non-compatible Systems

Three-parameter Regularization Algorithm for Pseudo-solution of Non-compatible Systems

Deep Learning Techniques for Spatiotemporal Weighted Pose Taekwondo Features and Their Application in Tactical Analysis

Taekwondo behavior recognition has become a popular study issue in the past few decades due to its vast range of applications in the visual realm. The research of Taekwondo behavior recognition based on skeleton sequences has received increasing attention in recent years due to the widespread use of depth sensors and the development of real-time skeleton estimate methods based on depth images. In order to characterize the behavioral sequences, the majority of research work currently in existence extracts the spatial domain information of various skeleton joints within frames and the temporal domain information of the skeleton joints between frames. However, this research work ignores the fact that different joints and postures play different roles in determining the behavioral categories. Consequently, this paper presents a spatio-temporal weighted gesture Taekwondo features-based approach for Taekwondo recognition that employs a bilinear classifier to iteratively compute the weights of the static gestures and joint points relative to the action category in order to identify the joint points and gestures with high information content; concurrently, this paper introduces dynamic temporal regularization and Fourier time pyramid algorithms for temporal modeling in order to provide a better temporal analysis of the behavioural features, and ultimately employs support vector machines to complete the behavioural classification. According to experimental results on several datasets, this strategy outperforms certain other methods in terms of recognition accuracy and is highly competitive.

Sparse smooth group [formula omitted] regularization method for convolutional neural networks

Deep convolutional neural networks (CNNs) have successfully addressed numerous challenging real-world problems and are known to be more complex and powerful classifiers. However, finding a simple CNN architecture to properly estimate an unknown solution is still a challenging problem. With the objective of regularizing networks at the group level instead of individual ones, we propose the sparse smooth Group L0∘L1/2 regularization method, which is formulated as a composition of L0 and L1/2 regularizers. In doing so, both the L0 and L1/2 regularization terms are nonsmooth, which leads to difficulties in the training process. To overcome this issue, we propose a novel smooth function approximating the L1/2 regularizer. Additionally, we use an existing smooth indicator function to approximate the L0 regularizer. Numerical simulation results on a range of benchmark datasets and various CNN architectures demonstrate the effectiveness of the proposed method. It outperforms well-known group sparse regularization algorithms in several aspects, particularly in achieving superior group and weight sparsity.

Estimating PM2.5 utilizing multiple linear regression and ANN techniques

The accurate prediction of air pollutants, particularly Particulate Matter (PM), is critical to support effective and persuasive air quality management. Numerous variables influence the prediction of PM, and it's crucial to combine the most relevant input variables to ensure the most dependable predictions. This study aims to address this issue by utilizing correlation coefficients to select the most pertinent input and output variables for an air pollution model. In this work, PM2.5 concentration is estimated by employing concentrations of sulfur dioxide, nitrogen dioxide, and PM10 found in the air through the application of Artificial Neural Networks (ANNs). The proposed approach involves the comparison of three ANN models: one trained with the Levenberg–Marquardt algorithm (LM-ANN), another with the Bayesian Regularization algorithm (BR-ANN), and a third with the Scaled Conjugate Gradient algorithm (SCG-ANN). The findings revealed that the LM-ANN model outperforms the other two models and even surpasses the Multiple Linear Regression method. The LM-ANN model yields a higher R2 value of 0.8164 and a lower RMSE value of 9.5223.