Step Size Research Articles

Ensemble Transfer Learning for Image Classification

The deep learning (DL) techniques used for image classification might not deliver the desired level of classification accuracy as some features belonging to some class of a dataset are missed during feature extraction. The ensemble learning (EL) based model improves classification accuracy by combining the strengths of individual classifiers. As a result, those features that were missed during feature extraction by a specific DL technique will be taken care of by another DL technique in an ensemble DL approach. In this paper, averaging EL (AENet), weighted averaging EL (WAENet), and stacking EL (StackedNet) approaches are proposed, considering the DenseNet201, EcientNetB0, and ResNetRS101 as base models. The predictions of the base models are averaged to generate the AENet. The WAENet is constructed by assigning weights to each base model based on their prediction and then taking their average. Similarly, the Stacked-Net is developed by considering the DenseNet201, EcientNetB0, and ResNetRS101 as base-learners and ResNetRS101 as meta-learner. Analysed performance of the considered pre-trained base models and the developed EL models on the standard and application-specific datasets such as MiniImageNet, CIFAR10, CIFAR100, Plant Village (PV), Tomato, Covid-19 and 9IndianFood. 80% of the datasets were used to train and 20% to test the base and proposed models. The models are trained for an epoch of 30, considering a learning rate of 0.001 and adam optimizer. The stackedNet delivered better results than others.

Stochastic Gradient Descent with Pre-Conditioned Polyak Step-Size

Stochastic Gradient Descent (SGD) is one of the many iterative optimization methods that are widely used in solving machine learning problems. These methods display valuable properties and attract researchers and industrial machine learning engineers with their simplicity. However, one of the weaknesses of this type of methods is the necessity to tune learning rate (step-size) for every loss function and dataset combination to solve an optimization problem and get an efficient performance in a given time budget. Stochastic Gradient Descent with Polyak Step-size (SPS) is a method that offers an update rule that alleviates the need of fine-tuning the learning rate of an optimizer. In this paper, we propose an extension of SPS that employs preconditioning techniques, such as Hutchinson’s method, Adam, and AdaGrad, to improve its performance on badly scaled and/or ill-conditioned datasets.

A Non-Standard Finite Difference Discretization Scheme Applied to a Malaria Model

In this research work, a dynamically consistent non-standard finite difference (NSFD) scheme is developed to solve a continuous-time model of malaria transmission with herbal medicine as control strategy. We compared results from NSFD scheme with the standard finite difference methods (4th order Runge-kutta and forward Euler methods). The numerical investigation showed that the proposed NSFD method remains consistent, preserves the positivity of solutions and converges to true equilibrium points of the continuous model independent of the step size h.

Higher-order iterative decoupling for poroelasticity

AbstractFor the iterative decoupling of elliptic–parabolic problems such as poroelasticity, we introduce time discretization schemes up to order five based on the backward differentiation formulae. Its analysis combines techniques known from fixed-point iterations with the convergence analysis of the temporal discretization. As the main result, we show that the convergence depends on the interplay between the time step size and the parameters for the contraction of the iterative scheme. Moreover, this connection is quantified explicitly, which allows for balancing the single error components. Several numerical experiments illustrate and validate the theoretical results, including a three-dimensional example from biomechanics.

An intelligent magnetic resonance imagining-based multistage Alzheimer's disease classification using swish-convolutional neural networks.

Alzheimer's disease (AD) refers to a neurological disorder that causes damage to brain cells and results in decreasing cognitive abilities and memory. In brain scans, this degeneration can be seen in different ways. The disease can be classified into four stages: Non-demented (ND), moderate demented (MoD), mild demented (MiD), and very mild demented (VMD). To prepare the raw dataset for analysis, the collected magnetic resonance imaging (MRI) images are subjected to several pre-processing techniques in order to improve the performance accuracy of the proposed model. Medical images generally have poor contrast and get affected by noise, which ends up with inaccurate diagnosis. For the different phases of AD to be detected, a clear image is necessary. To address this issue, the influence of the artefacts must be reduced, enhance the contrast, and reduce the loss of information. A novel framework for image enhancement is suggested to increase the accuracy in the detection and identification of AD. In this study, the raw MRI dataset from the Alzheimer's disease neuroimaging initiative (ADNI) database is subjected to skull stripping, contrast enhancement, and image filtering followed by data augmentation to balance the dataset with four types of Alzheimer's classes. The pre-processed data are subjected to five different pre-trained models such as AlexNet, ResNet, VGG 16, EfficientNet, and Inceptionv3 achieving a testing accuracy rate of 91.2%, 88.21%, 92.34%, 93.45%, and 85.12%, respectively. These pre-trained models are compared with the proposed CNN (convolutional neural network) model designed with Adam optimizer and Flatten Swish activation function which reaches the highest accuracy of 96.5% with a learning rate of 0.000001. The five pre-trained CNNmodels along with the proposed swish-based AD-CNN were tested using various performance metrics to evaluate the model efficiency in classifying and identifying the AD classes. From the result analysis, it is evident that the proposed AD-CNN model outperforms all the other models.

Steepest descent method for uncertain multiobjective optimization problems under finite uncertainty set

In this paper, a steepest descent method is developed for the robust counterpart of uncertain multiobjective optimization problems under finite uncertainty. The robust counterpart under consideration is the minimum of objective wise worst case, which is a nonsmooth deterministic multiobjective optimization problem. To solve this robust counterpart with the help of the steepest descent method, a subproblem is constructed and solved to find a descent direction. An Armijo-type inexact line search technique is employed to find a suitable step length. With the help of the descent direction and step length, a sequence is generated. The convergence of the proposed method is justified under some common assumptions. Finally, the algorithm is verified and compared with one existing method with the help of some numerical problems.

A Coverage Hole Recovery Method for 3D UWSNs Based on Virtual Force and Energy Balance

Underwater wireless sensor networks (UWSNs) have been applied in lots of fields. However, coverage holes are usually caused by complex underwater environment. Coverage holes seriously affect UWSNs’ performance and quality of service; thus, their recovery is crucial for 3D UWSNs. Although most of the current research recovery algorithms demand hole detection, the number of additional mobile nodes is too large, the communication and computing costs are high, and the coverage and energy balance are poor. Therefore, these methods are not suitable for UWSN hole repairing. In order to enhance the performance of hole recovery, a coverage hole recovery method for 3D UWSNs in complex underwater environments based on virtual force guidance and energy balance is proposed. The proposed method closely combines the node energy and considers complex environmental factors. A series of multi-dimensional virtual force models are established based on energy between nodes, area boundaries, zero-energy holes, low-energy coverage holes, underwater terrain, and obstacle forces. Then, a coverage hole recovery method for 3D UWSNs based on virtual force guidance and energy balance (CHRVE) is proposed. In this method, the direction and step size of mobile repairing node movement is guided by distributed computation of virtual forces, and the nodes are driven towards the target location by means of AUV or other carrier devices. The optimal position to improve coverage rate and node force balance is obtained. Simulation experiments show good adaptability and robustness to complex underwater terrain and different environments. The algorithm does not require precise coverage hole boundary detection. Furthermore, it balances network energy distribution significantly. Therefore, this method reduces the frequency of coverage hole emergence and network maintenance costs.

Event‐Triggered Adaptive Neural Network Control for 4WS4WD Wheeled Mobile Robot

ABSTRACTIn this article, an event‐triggered adaptive neural network controller based on threshold band is designed for a four wheels independently steered and four wheels independently driven (4WS4WD) mobile robot. The 4WS4WD mobile robot is attracting attention for its excellent motion performance such as manipulation versatility and posture flexibility. However, its control difficulty is increased due to its characteristics of being controlled by eight motors for steering and driving respectively. Also, since the robot itself has limited computing and communication resources, real‐time control cannot be guaranteed. To copy with that, the kinematic and dynamics models are first introduced for the 4WS4WD mobile robot. Second, an adaptive neural network controller with low‐frequency learning rate is utilized to control the mobile robot since there are unknown perturbations in the model. It can maintain system stability while handing unidentified model perturbations. The stability of the controller is demonstrated by Lyapunov stability analysis. An event‐triggered based on threshold band is suggested to lessen the amount of computation in the control process. Finally, the simulation outcomes further demonstrate how the suggested approach can greatly lessen the computational and communication cost while maintaining control performance.

Analysis of the accuracy of the inverse marching method used to determine thermal stresses in cylindrical pressure components

This paper analyses the inverse marching method used to determine the thermal stresses on the inner surface of a thick-walled cylindrical element not weakened by holes in the transient state. The heat conduction problem was considered one-dimensional, i.e. it was assumed that heat is transferred only in the radial direction. The method is based on measuring the temperature inside the pipeline wall at a single point and assuming that the pipeline is thermally insulated. The paper undertook an evaluation of the influence of the measuring point's distance from the inner surface, the number of control volumes into which the inverted area was divided and the length of the time step on the accuracy of the calculated temper-ature, heat transfer coefficient and thermal stresses on the inner surface of the pressure element. Verification was performed by comparing the calculation results obtained from the direct analytical method perturbed by random errors with those obtained from the numerical inverse step method.

Paving the way for future advancements in superconductivity research through gold ormus studies

Abstract Background Gold ormus is a type of superconductor that can exhibit superconductivity at temperatures below 1 Kelvin, allowing it to conduct electricity without resistance. While not as widely used as other materials like niobium or lead, gold ormus is valuable for research in superconductivity. Limited studies have been conducted on gold ormus. Numerical simulations of the Ginzburg–Landau theory have yielded important results for both gold ormus. Results Class-I and class-II superconducting gold ormus, have been successfully simulated using the Runge–Kutta fourth-order method. Our analysis shows the convergence of our simulation outcomes and emphasizes the importance of considering truncation error and selecting appropriate step sizes for accurate results. The periodic factor of penetration (PFP) for each superconductor has been determined, with class-I superconducting gold ormus having a PFP of 250 nm, class-II superconducting gold ormus having a PFP of 566.2 nm. The relationship between the PFP and the length of the penetration depth has also been revealed. Conclusions Our study confirms the accuracy of the Runge–Kutta fourth-order method in simulating superconductors. By analyzing the PFP for different superconducting materials, we have identified trends in penetration depth that contribute to understanding superconductivity. Our simulations provide valuable insights for further research in the field of superconductivity. Adjusting parameters carefully ensures reliable simulations and advances progress in superconductivity research.

Compact ADI Difference Scheme for the 2D Time Fractional Nonlinear Schrödinger Equation

In this paper, we will introduce a compact alternating direction implicit (ADI) difference scheme for solving the two-dimensional (2D) time fractional nonlinear Schrödinger equation. The difference scheme is constructed by using the L1−2−3 formula to approximate the Caputo fractional derivative in time and the fourth-order compact difference scheme is adopted in the space direction. The proposed difference scheme with a convergence accuracy of O(τ1+α+hx4+hy4)(α∈(0,1)) is obtained by adding a small term, where τ, hx, hy are the temporal and spatial step sizes, respectively. The convergence and unconditional stability of the difference scheme are obtained. Moreover, numerical experiments are given to verify the accuracy and efficiency of the difference scheme.

Superconvergence analysis of a low order expanded conforming mixed finite element method for the extended Fisher–Kolmogorov equation

This paper is devoted to the superconvergence analysis of a low order expanded conforming mixed finite element method(MFEM) for the extended Fisher–Kolmogorov equation. Both semi-discrete and linearized backward Euler fully-discrete schemes are analyzed with bilinear and Nédéleć elements. The present work has two main contributions: One is that some a priori bounds of the approximations are derived, which lead to the unique solvability of the semi-discrete scheme and make the nonlinear term can be dealt with efficiently. By such a way, the superclose properties and superconvergence results can be derived through the high accuracy results of the above two elements and the interpolated postprocessing technique, respectively. The other one is that a novel splitting technique is utilized to deal with the nonlinear term, which can avoid the use of popular error splitting technique and the restrictions between the time step size and mesh size. Besides, the combination technique of interpolation and projection operators also plays an important role in the superclose and superconvergence analysis. Finally, some numerical results are provided to show the validity of the theoretical analysis.

Abstract 4139194: Predicting Cholesterol Screening Behavior After Age 50 Using Machine Learning: Insights from the Health and Retirement Study

Background: In the U.S., about 8% of adults never received cholesterol screening. Although machine learning (ML) has been used to develop decision tools for Atherosclerotic Cardiovascular Disease (ASCVD) risk prediction, its application in behavioral forecasting has not yet been explored in the context of cholesterol screening behaviors. This study aimed to examine the performance and accuracy of ML algorithms in forecasting cholesterol screening behaviors in adults after age 50. Methods: This analysis used deidentified data from the Health and Retirement Study (HRS) 2004-2018. HRS is a longitudinal survey among 23,000 households in the U.S. Participants were excluded from the current analysis if they passed away by 2019, ever had ASCVD or stroke, were under age 50 at baseline, or had missing data in self-reported cholesterol screening. In total, 7176 participants (mean age [SD]=62 [8]) met the inclusion criteria; participants were randomly split into a training set (80%) and a testing set (20%). The synthetic minority oversampling technique was used to solve the imbalance distribution of the rare event. Five ML algorithms were used: random forest, gradient boosting machine (GBM), XGBoost, Support Vector Machine (SVM), and logistic regression. Accuracy, AUROC, and positive predictive value (PPV) were used to compare model performance. The average gain was evaluated for feature importance in the demographic and health domains. Results: In total, 232 (3.2%) respondents did not receive any cholesterol screening from 2008 to 2018. Experiments with five ML algorithms suggested that XGBoost with deeper trees and learning rate performed better in classifying those who did not screen for cholesterol levels over 10 years. Adding prior cholesterol screening history (2004-2006) into the model significantly improved model performance. Hypertension, self-rated health, and smoking were the major health features, while insurance, poverty, and work status were the major demographic features in the predictive model (accuracy=0.97; AUROC=0.88; PPV=0.42). Conclusion: Findings underscore the potential utility of ML models in predicting cholesterol screening behaviors after age 50. This could be the basis for developing decision tools for clinicians to identify those with a lower chance of cholesterol screening or make reminders accordingly. The low-cost predictive model might improve the uptake of preventive screening behaviors in middle-aged and older adults.

Intelligent road surface state recognition method based on multi-layer attention residual network

Abstract Data-driven road surface state recognition enhances the efficiency and accuracy of road management, contributing to increased safety and reliability in road traffic. However, traditional machine learning and deep learning-based road surface state recognition typically rely on extensive data for model training, making it challenging to adapt to complex tasks in diverse scenarios. Therefore, this paper proposes a Multi-layer Attention Residual Network (MARN)-based intelligent road surface state recognition method. First, a Residual Convolutional Neural Network (ResNet) is constructed as the backbone model of MARN to mitigate the gradient vanishing problem, allowing the network to extract deeper features. Subsequently, an adaptive multi-layer attention mechanism is introduced in each convolutional layer, enabling adaptive weighting of each feature channel in the dataset to enhance the model’s focus on different features for better feature extraction. Furthermore, a cosine annealing learning rate adjuster is designed to improve the accuracy, robustness, and convergence during the model training process. Finally, the proposed MARN is validated using an image dataset containing six different road surface states. Comparative studies are conducted on the recognition accuracy of the proposed MARN, original ResNet, Visual Geometry Group network (VGG16), and Convolutional Neural Network (CNN). The impact of different batch sizes on the convergence speed of road surface state recognition under MARN is also analyzed. Results demonstrate that MARN achieves a training set accuracy of over 95%, surpassing VGG16 and CNN with accuracies below 85%. Compared to ResNet, MARN exhibits a 1.3% higher training set accuracy and a 0.25 lower validation set loss, showcasing superior accuracy and robustness in road surface state recognition.

Image–Text Matching Model Based on CLIP Bimodal Encoding

Image–text matching is a fundamental task in the multimodal research field, connecting computer vision and natural language processing by aligning visual content with corresponding textual descriptions. Accurate matching is critical for applications such as image captioning and text-based image retrieval yet remains challenging due to the differences in data modalities. This paper addresses these challenges by proposing a robust image–text matching model inspired by Contrastive Language–Image Pre-training (CLIP). Our approach employs the Vision Transformer (ViT) model as the image encoder and Bidirectional Encoder Representations from Transformers (Bert) as the text encoder, integrating these into a shared vector space to measure semantic similarity. We enhance the model’s training efficiency using the LiT-tuning paradigm to optimize learning through a cosine decay strategy for dynamic adjustment of the learning rate. We validate our method on two benchmark datasets, WuKong and Flickr30k, demonstrating that our model achieves superior performance and significantly improves key evaluation metrics. The results underscore the model’s effectiveness in achieving accurate and robust image–text alignment.

Ensuring Driving and Road Safety of Autonomous Vehicles Using a Control Optimiser Interaction Framework Through Smart “Thing” Information Sensing and Actuation

Road safety through point-to-point interaction autonomous vehicles (AVs) assimilate different communication technologies for reliable and persistent information sharing. Vehicle interaction resilience and consistency require novel sharing knowledge for retaining driving and pedestrian safety. This article proposes a control optimiser interaction framework (COIF) for organising information transmission between the AV and interacting “Thing”. The framework relies on the neuro-batch learning algorithm to improve the consistency measure’s adaptability with the interacting “Things”. In the information-sharing process, the maximum extraction and utilisation are computed to track the AV with precise environmental knowledge. The interactions are batched with the type of traffic information obtained, such as population, accidents, objects, hindrances, etc. Throughout travel, the vehicle’s learning rate and the surrounding environment’s familiarity with it are classified. The learning neurons are connected to the information actuated and sensed by the AV to identify any unsafe vehicle activity in unknown or unidentified scenarios. Based on the risk and driving parameters, the safe and unsafe activity of the vehicles is categorised with a precise learning rate. Therefore, minor changes in vehicular decisions are monitored, and driving control is optimised accordingly to retain 7.93% of navigation assistance through a 9.76% high learning rate for different intervals.

Foot Progression Angle in Relation to Spatiotemporal Parameters of Gait in Children with Cerebral Palsy

OBJECTIVE: This study assessed the difference in the foot progression angle and measure spatiotemporal parameters of gait of children with diplegic cerebral palsy and to examine the relation among foot progression angle of right and left lower limb and spatiotemporal parameters of gait. DESIGN: Cross-sectional and Correlation study. SUBJECTS: Sixty children with spastic diplegia, aged 5 to 8 years. The participants were categorized into 3 age groups of equivalent number: group A (5 to 6 years), group B (6 to 7 years), as well as group C (7 to 8 years). METHODS: Foot progression angle and spatiotemporal parameters of the gait for both feet were evaluated utilizing a dynamic footprint. RESULTS: a moderate negative significant correlation was noted among left FPA and right step length, and right stride length. There was moderate positive significant correlation among left FPA and cadence . A weak non-significant correlation was observed among right and left FPA and gait parameters among three groups. CONCLUSION: foot progression angle assessment and spatiotemporal measurements and the correlation among them can give objective and quantitative data that could be utilized in rehabilitation and clinical evaluations to identify functional deficiencies.

Heuristic strategy using hybrid deep learning with transfer learning for oral cancer detection

Oral cancer becomes the most disastrous ailment that affects the oral cavity parts of the mouth. Oral cancer diagnosis is the main challenge in the medical field. It becomes expensive and less capable of classifying oral cancer. In some cases, it may cause unnecessary morbidity and mortality. Recently, the detection of malignant and premalignant oral lesions has been a critical process owing to their low image resolution and lower acquisition time. Thus, a novel hybrid deep learning with meta-heuristic-based optimization is proposed. The pre-processing occurs by median filtering and Contrast Limited Adaptive Histogram Equalization (CLAHE). The CLAHE method is utilized to reduce unwanted noise. Finally, the classification is done by a proposed hybrid-based deep learning model termed as Recurrent Deep Belief Network (RDBN), in which the Deep Belief Network (DBN) is incorporated with the Recurrent Neural Network (RNN). Here, the RDBN helps to increase the performance classification. Furthermore, the hyperparameters of the RDBN model, such as learning rate, epochs and hidden neurons, are tuned using the proposed Hybrid Beetle-Barnacle Swarm Optimization (HBBSO) algorithm, where the Barnacles Mating Optimizer (BMO) is superimposed with Beetle Swarm Optimization (BSO) algorithms. In the given proposed model, the selected features are extracted by the optimization algorithms. Here, the parameters are fine-tuned to get the better optimal solution. From the experimental outcome, the developed model has acquired 5.2% better than PSO-RDBN, 5.6% improved than GWO-RDBN, 3.2% enhanced than BSO-RDBN and 3.5% superior to BMO-RDBN regarding accuracy. Thus, the proposed model achieves higher results for detecting oral cancer with enhanced classification performance than the existing approaches.

A medical disease assisted diagnosis method based on lightweight fuzzy SZGWO-ELM neural network model.

The application of neural network model in intelligent diagnosis usually encounters challenges such as continuous adjustment of network parameters and significant cost in training the network facing numerous complex physiological data. To address this challenge, this paper introduces a fuzzy SZGWO-ELM neural network model for medical disease aid diagnosis with fuzzy membership function and ELM network to refine the improved Gray Wolf optimization algorithm. Firstly, the Z-type membership function is introduced as the inertia weight to get a balance for the grey wolf in seeking the optimal solution globally and locally and ensuring fast convergence. Secondly, the S-type membership function is utilized as the adaptive weight to flexibly adjust the grey wolf search step size to facilitate a quick approximation of the optimal solution. Finally, the improved Gray Wolf optimization algorithm is used to optimize the parameters of the ELM neural network model, termed as SZGWO-ELM. This method can eliminate the need for extensive network parameter adjustments and quickly locate the optimal solution to the problem using a lightweight neural network. The performance of the SZGWO is assessed by using metrics like convergence, mean, and standard deviation. Multiple experiments reveal that this method shows superior performance. Furthermore, five publicly accessible medical disease datasets from UCI were conducted to evaluate the performance of SZGWO-ELM network model comparing with different classify model, and the results in terms of precision, sensitivity, specificity and accuracy can achieve 99.52%, 94.14%, 99.26% and 96.08%, respectively, which illustrate that the proposed SZGWO-ELM neural network significantly enhance the model's accuracy, providing better support for doctors in disease diagnosis.

NIMG-82. DEVELOPMENT AND VALIDATION OF AN MRI-BASED DEEP LEARNING MODEL TO DIFFERENTIATEIDH-WILDTYPE GLIOBLASTOMA AND TUMEFACTIVE MULTIPLE SCLEROSIS

Abstract Clinical management of IDH wildtype glioblastoma (GBM) and tumefactive multiple sclerosis (tMS) is drastically different. GBM requires maximal safe resection followed by chemoradiation, while tMS outcome is worsened by surgery and radiotherapy. Noninvasive methods are needed to help with accurate diagnosis of tumor and non-tumor etiologies. To develop an MRI-based classification model, tMS subjects diagnosed prior to January 1, 2020, were matched to tMS by age at diagnosis, sex, index MRI date, and 2D/3D acquisition. Inclusion criteria included one cm minimal lesion size and pre-operative post-contrast T1 and T2 images available for analysis. A 3D-DenseNet121 was used to develop a classification model using prespecified parameters: 650 epochs, batch size 16, learning rate 10-3, cross-entropy loss, and AdamW optimizer. The stopping rule was defined as three sequential differences in epoch cross-entropy loss <0.02. Models were developed using both T1gd and T2, as well as from only T1gd and only T2. Training included 220 subjects (110 GBM, 110 tMS). A 2-stage validation design was used, which included both retrospective and prospective cohorts. Stage 1 consisted of 272 retrospective GBM (diagnosed prior to January 1, 2020). Stage 2 consisted of 69 and 34 prospective (diagnosed after January 1, 2020) GBM and tMS, respectively. External validation on the 272 retrospective GBM demonstrated accuracy of 91%, 84%, and 78% for T1gd+T2, T1gd only, and T2 only, respectively. External validation on the 69 prospective GBM demonstrated an accuracy of 87%, 64%, and 67% for T1gd+T2, T1gd only, and T2 only, respectively. The 34 prospective tMS demonstrated accuracy of 76%, 76%, and 82% for T1gd+T2, T1gd only, and T2 only, respectively. This shows the feasibility of deep learning to aid in differential diagnosis of brain lesions. Future work entails the integration of germline variants into the classification model, including variants associated with the risk of glioma or MS.