Maximum Prediction Accuracy Research Articles

Analysis of Local Structure of Mechanical and Thermal Rearrangements in Glasses with the Atomic Cluster Expansion.

We explore the structural signatures of excitations in amorphous materials with the atomic cluster expansion (ACE), a universal and complete linear basis of descriptors of the atomic environment. Body-orderd linear classifiers are constructed that distinguish between active and inactive particles in three different model glass formers, in which structural relaxation occurs either through spontaneous thermal activation or by simple shear. We find that in binary mixtures, maximum prediction accuracy is already achieved with very few two-body correlations, while a polymer glass requires both two- and three-body correlations. Trends are robust across both activation mechanisms.

GREMI: An Explainable Multi-Omics Integration Framework for Enhanced Disease Prediction and Module Identification.

Multi-omics integration has demonstrated promising performance in complex disease prediction. However, existing research typically focuses on maximizing prediction accuracy, while often neglecting the essential task of discovering meaningful biomarkers. This issue is particularly important in biomedicine, as molecules often interact rather than function individually to influence disease outcomes. To this end, we propose a two-phase framework named GREMI to assist multi-omics classification and explanation. In the prediction phase, we propose to improve prediction performance by employing a graph attention architecture on sample-wise co-functional networks to incorporate biomolecular interaction information for enhanced feature representation, followed by the integration of a joint-late mixed strategy and the true-class-probability block to adaptively evaluate classification confidence at both feature and omics levels. In the interpretation phase, we propose a multi-view approach to explain disease outcomes from the interaction module perspective, providing a more intuitive understanding and biomedical rationale. We incorporate Monte Carlo tree search (MCTS) to explore local-view subgraphs and pinpoint modules that highly contribute to disease characterization from the global-view. Extensive experiments demonstrate that the proposed framework outperforms state-of-the-art methods in seven different classification tasks, and our model effectively addresses data mutual interference when the number of omics types increases. We further illustrate the functional- and disease-relevance of the identified modules, as well as validate the classification performance of discovered modules using an independent cohort.

Constructing a visual detection method for coagulation effect based on image feature machine learning

The coagulation process is influenced by several factors, including turbidity, temperature, pH, and hydraulic conditions. Consequently, optimizing the dosage of coagulants is often time-consuming and challenging. To quickly evaluate coagulation efficacy and optimize coagulant dosage, this study utilized a convolutional neural network (CNN) model to analyze the accuracy of effluent turbidity predictions from floc images at different time intervals. Python OpenCV was employed to develop a program that extracts macro features (e.g., texture) and micro features (e.g., particle size distribution) from the images, and these feature parameters were analyzed using a BP-ANN model. The results of the CNN analysis indicated that the middle to late stages of flocculation are the optimal periods for predicting effluent turbidity, with the highest prediction accuracy of 99.81 % achieved during the middle stage of flocculation. The BP-ANN analysis demonstrated that particle size distribution effectively represents the microscopic characteristics of flocs, achieving a maximum prediction accuracy of 96.94 % in the middle stage of flocculation. Furthermore, combining macroscopic and microscopic features yielded a prediction accuracy of 99.44 % at the end of flocculation using BP-ANN. These findings suggest that machine learning techniques applied to floc images can effectively predict effluent turbidity, offering valuable insights for future water quality prediction and the development of flocculation kinetics models.

Story Recall in Peer Conflict Resolution Discourse Task to Identify Older Adults Testing Within Range of Cognitive Impairment.

The current study used behavioral measures of discourse complexity and story recall accuracy in an expository discourse task to distinguish older adults testing within range of cognitive impairment according to a standardized cognitive screening tool in a sample of self-reported healthy older adults. Seventy-three older adults who self-identified as healthy completed an expository discourse task and neuropsychological screener. Discourse data were used to classify participants testing within range of cognitive impairment using multiple machine learning algorithms and stability analysis for identifying reliably predictive features in an effort to maximize prediction accuracy. We hypothesized that a higher rate of pronoun use and lower scores on story recall would best classify older adults testing within range of cognitive impairment. The highest classification accuracy exploited a single variable in a remarkably intuitive way: using 66% story recall as a cutoff for cognitive impairment. Forcing this decision tree model to use more features or increasing its complexity did not improve accuracy. Permutation testing confirmed that the 77% accuracy and 0.18 Brier skill score achieved by the model were statistically significant (p < .00001). These results suggest that expository discourse tasks that place demands on executive functions, such as working memory, can be used to identify aging adults who test within range of cognitive impairment. Accurate representation of story elements in working memory is critical for coherent discourse. Our simple yet highly accurate predictive model of expository discourse provides a promising assessment for easy identification of cognitive impairment in older adults. https://doi.org/10.23641/asha.26543824.

Assessment of Envelope- and Machine Learning-Based Electrical Fault Type Detection Algorithms for Electrical Distribution Grids

This study introduces envelope- and machine learning (ML)-based electrical fault type detection algorithms for electrical distribution grids, advancing beyond traditional logic-based methods. The proposed detection model involves three stages: anomaly area detection, ML-based fault presence detection, and ML-based fault type detection. Initially, an envelope-based detector identifying the anomaly region was improved to handle noisier power grid signals from meters. The second stage acts as a switch, detecting the presence of a fault among four classes: normal, motor, switching, and fault. Finally, if a fault is detected, the third stage identifies specific fault types. This study explored various feature extraction methods and evaluated different ML algorithms to maximize prediction accuracy. The performance of the proposed algorithms is tested in an emulated software–hardware electrical grid testbed using different sample rate meters/relays, such as SEL735, SEL421, SEL734, SEL700GT, and SEL351S near and far from an inverter-based photovoltaic array farm. The performance outcomes demonstrate the proposed model’s robustness and accuracy under realistic conditions.

Integrated error modeling for abrasive waterjet cutting at various nozzle tilt angle

The research investigates the specific impact of high-pressure abrasive waterjet (AWJ) on the cutting accuracy of circular trajectories. The coupling relationship between the kerf taper and the jet lag of the cutting front profiles is analyzed. Thus, an integrated error model for abrasive waterjet cutting of circular trajectories is innovatively established, which can predict the deformation of the circular arc trajectory not only at right angles but also at tilt angles. Cutting experiments at vertical angles were conducted, with a maximum prediction accuracy of 2.41%. Besides, arc trajectories with different diameters were conducted with a prediction accuracy of 2.26%. In order to reduce the deformation error of arc trajectories, compensation experiments for nozzle tilt ranging from 76° to 94° were carried out. The experimental results showed that when the nozzle was tilted in the direction opposite to the traverse speed up to half of the maximum deviation angle at the jet outlet point, the maximum deformation of arc trajectories was 1.218 mm, which was reduced by 0.532 mm compared to the case of vertical incidence, and the cylindricity increased by 31.74%. This adjustment can meet the precision requirements for cutting thick materials with abrasive waterjet.

Multi-omics-based Machine Learning for the Subtype Classification of Breast Cancer

AbstractCancer is a complicated disease that produces deregulatory changes in cellular activities (such as proteins). Data from these levels must be integrated into multi-omics analyses to better understand cancer and its progression. Deep learning approaches have recently helped with multi-omics analysis of cancer data. Breast cancer is a prevalent form of cancer among women, resulting from a multitude of clinical, lifestyle, social, and economic factors. The goal of this study was to predict breast cancer using several machine learning methods. We applied the architecture for mono-omics data analysis of the Cancer Genome Atlas Breast Cancer datasets in our analytical investigation. The following classifiers were used: random forest, partial least squares, Naive Bayes, decision trees, neural networks, and Lasso regularization. They were used and evaluated using the area under the curve metric. The random forest classifier and the Lasso regularization classifier achieved the highest area under the curve values of 0.99 each. These areas under the curve values were obtained using the mono-omics data employed in this investigation. The random forest and Lasso regularization classifiers achieved the maximum prediction accuracy, showing that they are appropriate for this problem. For all mono-omics classification models used in this paper, random forest and Lasso regression offer the best results for all metrics (precision, recall, and F1 score). The integration of various risk factors in breast cancer prediction modeling can aid in early diagnosis and treatment, utilizing data collection, storage, and intelligent systems for disease management. The integration of diverse risk factors in breast cancer prediction modeling holds promise for early diagnosis and treatment. Leveraging data collection, storage, and intelligent systems can further enhance disease management strategies, ultimately contributing to improved patient outcomes.

Automatic ploidy prediction and quality assessment of human blastocysts using time-lapse imaging

Assessing fertilized human embryos is crucial for in vitro fertilization, a task being revolutionized by artificial intelligence. Existing models used for embryo quality assessment and ploidy detection could be significantly improved by effectively utilizing time-lapse imaging to identify critical developmental time points for maximizing prediction accuracy. Addressing this, we develop and compare various embryo ploidy status prediction models across distinct embryo development stages. We present BELA, a state-of-the-art ploidy prediction model that surpasses previous image- and video-based models without necessitating input from embryologists. BELA uses multitask learning to predict quality scores that are thereafter used to predict ploidy status. By achieving an area under the receiver operating characteristic curve of 0.76 for discriminating between euploidy and aneuploidy embryos on the Weill Cornell dataset, BELA matches the performance of models trained on embryologists’ manual scores. While not a replacement for preimplantation genetic testing for aneuploidy, BELA exemplifies how such models can streamline the embryo evaluation process.

Assessing the transportability of radiomic models for lung cancer diagnosis: commercial vs. open-source feature extractors.

Radiomics has shown promise in improving malignancy risk stratification of indeterminate pulmonary nodules (IPNs) with many platforms available, but with no head-to-head comparisons. This study aimed to evaluate transportability of radiomic models across platforms by comparing performances of a commercial radiomic feature extractor (HealthMyne) with an open-source extractor (PyRadiomics) on diagnosis of lung cancer in IPNs. A commercial radiomic feature extractor was used to segment IPNs from computed tomography (CT) scans, and a previously validated radiomic model based on commercial features was used as baseline (ComRad). Using same segmentation masks, PyRadiomics, an open-source feature extractor was used to build three open-source radiomic models (OpenRad) using different methods: de novo open-source model derived using least absolute shrinkage and selection operator (LASSO) for feature selection, selecting open-source features matched to ComRad features based upon Imaging Biomarker Standardization Initiative (IBSI) nomenclature, and selecting open-source features most highly correlated to ComRad features. Radiomic models were trained on an internal cohort (n=161) and externally validated on 3 cohorts (n=278). We added Mayo clinical risk score to OpenRad and ComRad models, creating integrated clinical radiomic (ClinRad) models. All models were compared using area under the curve (AUC) and evaluated for clinical improvement using bias-corrected clinical net reclassification indices (cNRIs). ComRad AUC was 0.76 [95% confidence interval (CI): 0.71-0.82], and OpenRad AUC was 0.75 (95% CI: 0.69-0.81) for LASSO model, 0.74 (95% CI: 0.68-0.79) for Spearman's correlation, and 0.71 (95% CI: 0.65-0.77) for IBSI. Mayo scores were added to OpenRad LASSO model, which performed best, forming open-source ClinRad model with AUC of 0.80 (95% CI: 0.74-0.86), identical to commercial ClinRad's AUC. Both ClinRad models showed clinical improvement compared to Mayo alone, with commercial ClinRad achieving cNRI of 0.09 (95% CI: 0.02-0.15) for benign and 0.07 (95% CI: 0.00-0.13) for malignant, and open-source ClinRad achieving cNRI of 0.09 (95% CI: 0.02-0.15) for benign and 0.06 (95% CI: 0.00-0.12) for malignant. Transportability of radiomic models across platforms directly does not conserve performance, but radiomic platforms can provide equivalent results when building de novo models allowing for flexibility in feature selection to maximize prediction accuracy.

Evaluation of stress distributions in trimaterial bonded joints with nano-resin adhesive using machine learning models

Adhesive bonded joints hold significant importance across various industrial sectors in modern engineering, owing to their lightweight nature and myriad advantages. The rising demand for trimaterial joints underscores their utility and versatility. In these joints, the choice of materials for both adherends greatly influences their strength, structural reliability, and overall characteristics. While numerous researches have extensively analyzed stress distributions, their effects, and behaviors, many have relied on a one-factor-at-a-time approach, focusing solely on individual design variables' effects. However, recognizing the intricate interplay among various material combinations and their collective impact on overall performance, this study employs various types of White-box, Black-box, and Grey-box machine learning algorithms to identify an optimized ML model as well as predict stress distributions for any random combinations of upper and lower adherend materials. Dataset of total 178 random material combinations were utilized for the training phases with 5-fold cross validation and model tuning. However, the decision tree regressor emerged as the optimized model by comparing the quantitative metrics of accuracy benchmark as well as the prediction outcomes obtained through all the machine learning models. The maximum prediction accuracy attained was an impressive 99.97 %, while the minimum recorded was 89.74 %. This research aims to identify tailored machine learning model specifically for trimaterial bonded joints where nano layer of resin is utilized as the adhesive.

End-to-end reinforcement learning of Koopman models for economic nonlinear model predictive control

(Economic) nonlinear model predictive control ((e)NMPC) requires dynamic models that are sufficiently accurate and computationally tractable. Data-driven surrogate models for mechanistic models can reduce the computational burden of (e)NMPC; however, such models are typically trained by system identification for maximum prediction accuracy on simulation samples and perform suboptimally in (e)NMPC. We present a method for end-to-end reinforcement learning of Koopman surrogate models for optimal performance as part of (e)NMPC. We apply our method to two applications derived from an established nonlinear continuous stirred-tank reactor model. The controller performance is compared to that of (e)NMPCs utilizing models trained using system identification, and model-free neural network controllers trained using reinforcement learning. We show that the end-to-end trained models outperform those trained using system identification in (e)NMPC, and that, in contrast to the neural network controllers, the (e)NMPC controllers can react to changes in the control setting without retraining.

Genetically optimised SMOTE-based adversarial discriminative domain adaptation for rotor fault diagnosis at variable operating conditions

Abstract For safety, reliability, and uninterrupted output of gas turbines, aviation engines, power-generating equipment, pumps, gears, compressors etc., rotor mass imbalance must be detected and diagnosed to avoid catastrophic failure. Industry 4.0 relies on predictive digital maintenance and deep learning-based convolutional neural network (CNN), which predicts defects but fails if the operating conditions change. Research studies in various fields indicate that the domain shift issue occurs due to source and target samples being from different domains, which reduces prediction capability. Moreover, research studies are scarce in examining prediction capability under varying operating speeds for rotor mass imbalance. Hence, this research proposes the adversarial discriminative domain adaptation (ADDA) technique which predicts machine failures under various operational conditions. The efficacy of ADDA has been explored by introducing 1D-CNN as a source and a target encoder inside ADDA's architecture to take advantage of CNN's feature extraction capability. Further, this research effectively tackles CNN's inherent issues of overfitting and hyperparameters value selection. Furthermore, The real-world scenario has more healthy samples than fault condition samples, causing a multiclass imbalance in sample data, which affects the classification decision boundary and causes biased prediction. Hence, the proposed methodology first addresses the class imbalance through synthetic minority oversampling (SMOTE), then genetic algorithm optimizes 1D-CNN's hyperparameters, and the effective dropout layer positioning solves the overfitting. Finally, the deep learning-based SMOTE_ADDA_GO-1D-CNN decreases domain discrepancy with adversarial discriminative domain adaptation. The proposed methodology's efficacy has been explored through F1-Score, which is used as multiclass evaluation metrics, and it has been benchmarked against standard machine learning and deep learning algorithms. The test results of the proposed methodology surpassed all of them with maximum prediction accuracy. Thus, this study contributes to rotor mass imbalance detection and diagnosis for multiclass imbalanced data under varying operational conditions by successfully overcoming potential challenges during fault prediction.

Utility of PET for Nodal Staging in Subsolid Clinical Stage IA (T1 N0) Lung Adenocarcinoma

BackgroundPositron emission tomography (PET) is the standard of care for non-small cell lung cancer (NSCLC) clinical staging, but it may have limited utility in evaluating subsolid lung adenocarcinomas that can have relatively indolent behavior without hypermetabolic activity. MethodsThe sensitivity and specificity of PET for determining pathologic lymph node status and disease-free survival were assessed in patients operated on for cT1 N0 subsolid lung adenocarcinoma from January 2006 to June 2022 (at Stanford University School of Medicine, Stanford, CA). Patients with clinical or pathologic tumor size >30 mm, hilar or mediastinal lymph node size >1cm, and purely solid tumors were excluded. ResultsPET was available in 498 of 534 (93.2%) patients and more often was used in older patients with larger and more solid tumors. The overall pathologic lymph node–positive rate was 8.4% (45 of 534). PET specificity was 95.1%, but sensitivity was only 20.0%. A tumor diameter of 18.5 mm and a solid component percentage of 62.5% had the maximum predictive accuracy for pathologic lymph node positivity, with a 0% and 1.5% rate of pathologic and PET lymph node positivity, respectively, for tumors with values lower than those thresholds. There was no significant difference in 5-year disease-free survival between individuals who did and did not undergo PET scanning (76.6% vs 96.8%; P = .07). Conversely, 134 (26.9%) patients who underwent PET scanning had 171 incidentally detected hypermetabolic lesions unrelated to lung cancer, with only 13 of 134 (9.7%) patients identified as having non-NSCLC premalignant or malignant conditions requiring further therapy. ConclusionsPET scan use for subsolid lung adenocarcinoma has high specificity but limited sensitivity for predicting pathologic lymph node positivity. PET also has no association with disease-free survival and often detects clinically unimportant findings rather than changing lung cancer management, particularly for patients with smaller and less solid tumors.

A hybrid tool wear prediction model based on JDA

PurposeAiming at solving the problems of low prediction accuracy and poor generalization caused by the difference in tool wear data distribution and the fixation of single global model parameters, a hybrid prediction modeling method for tool wear based on joint distribution adaptation (JDA) is proposed.Design/methodology/approachFirstly, JDA is exploited to adapt the data features with different data distributions. Then, the adapted data features are identified by the KNN classifier. Finally, according to the tool state classification results, different regression prediction models are assigned to different wear stages to complete the whole tool wear prediction task.FindingsThe results of milling experiments show that the maximum prediction accuracy of this method is 95.13%, and it has good recognition accuracy and generalization performance. Through the application of the tool wear hybrid prediction modeling method, the prediction accuracy and generalization performance of the model are improved and the tool monitoring is realized.Originality/valueThe research results can provide solutions and a theoretical basis for the application of tool wear monitoring technology in practical industrial applications.

Adaptive Method for Feature Selection in the Machine Learning Context

Feature selection is a fundamental aspect of machine learning that is crucial for improving the accuracy and efficiency of models. It carefully analyzes the abundance of data to identify the most significant characteristics, hence improving the accuracy of predictions and minimizing the likelihood of model overfitting. This technique not only optimizes model training by reducing computational requirements, but also enhances the model's interpretability, resulting in more transparent and reliable predictions. The deliberate omission of unnecessary variables is a process of improving the model and also constitutes a crucial measure toward achieving more flexible and comprehensible results in machine learning. An analysis to assess the effectiveness of feature selection on regression models was conducted. The impact was measured using Mean Squared Error (MSE) metrics. A variety of regression algorithms were evaluated, and then feature selection techniques, including statistical and algorithmic methods, such as SelectKBest, PCA, and RFE with Linear Regression and Random Forest, were applied. After selecting the features, linear models demonstrated improvements in mean squared error (MSE), highlighting the value of removing unnecessary data. This study emphasizes the subtle impact of feature selection on model performance, calling for a tailored strategy to maximize prediction accuracy.

Prediction of subthalamic stimulation efficacy on isolated dystonia via support vector regression

IntroductionDeep brain stimulation (DBS) of subthalamic nucleus (STN) has been well-established and increasingly applied in patients with isolated dystonia. Nevertheless, the surgical efficacy varies among patients. This study aims to explore the factors affecting clinical outcomes of STN-DBS on isolated dystonia and establish a well-performed prediction model. MethodsIn this prospective study, thirty-two dystonia patients were recruited and received bilateral STN-DBS at our center. Their baseline characteristics and up to one-year follow-up outcomes were assessed. Implanted electrodes of each subject were reconstructed with their contact coordinates and activated volumes calculated. We explored correlations between distinct clinical characteristics and surgical efficacy. Those features were then trained for the model in outcome prediction via support vector regression (SVR) algorithm and testified through cross-validation. ResultsPatients demonstrated an average clinical improvement of 56 ± 25 % after STN-DBS, significantly affected by distinct symptom forms and activated volumes. The optimal targets and activated volumes were concentratedly located at the dorsal posterior region to STN. Most patients had a rapid response to STN-DBS, and their motor score improvement within one week was highly associated with long-term outcomes. The trained SVR model, contributed by distinct weights of features, could reach a maximum prediction accuracy with mean errors of 11 ± 7 %. ConclusionSTN-DBS demonstrated significant and rapid therapeutic effects in patients with isolated dystonia, by possibly affecting the pallidofugal fibers. Early improvement highly indicates the ultimate outcomes. SVR proves valid in outcome prediction. Patients with predominant phasic and generalized symptoms, shorter disease duration, and younger onset age may be more favorable to STN-DBS in the long run.

Unmanned aerial vehicles (UAVs)-based crop lodging susceptibility and seed yield assessment during different growth stages of rapeseed (Brassica napus)

Lodging is a great challenge in rapeseed production that significantly affects seed yield and quality. Prediction of lodging susceptibility and seed yield through unmanned aerial vehicles (UAVs)-based framework offers remarkable prospects for higher applicability in agriculture. This study aims to explore the possibility of using a UAV-based framework for predicting the stem and root lodging susceptibility (represented by safety factor, SFs and SFr respectively) and seed yield at different growth stages of rapeseed. The Red-Green-Blue (RGB) and multispectral (MS) images were captured during various growth stages by UAV platforms to calculate 16 vegetation indices (VIs). Furthermore, the relationships of these VIs with lodging susceptibility and seed yield were also established using multiple linear regression (MLR) and four machine learning methods (including random forest machine (RFR), support vector machine, artificial neural network, and K-nearest neighbors). The results revealed that MS-VIs provided a good estimation of seed yield, and stem and root lodging susceptibilities. Among the 16 VIs analyzed, MS-VI SR85 emerged as the best predictor for both seed yield and lodging susceptibility, as is evident by its highest importance scores. Furthermore, when RGB-VIs were coupled with MS-VIs, the R2 values for estimating seed yield, stem lodging and root lodging resistance were enhanced by 150%, 69.6% and 106%, respectively, in comparison with RGB-VIs. Similarly, the RFR provided a more accurate machine learning method for predicting seed yield and lodging susceptibility compared to the other three models. Stem elongation stage was the optimum growth stage for the estimation of seed yield, and stem and root lodging susceptibilities due to its maximum prediction accuracy, as is suggested by the highest R2 values. It can be inferred that a UAV-based framework in combination with RFR could serve as a high-throughput technique for large-scale prediction of lodging susceptibility and seed yield, as early as at stem elongation stage and thus provides an opportunity for timely agronomic intervention.

Credit risk prediction based on an interpretable three-way decision method: Evidence from Chinese SMEs

Credit risk prediction can provide essential tools for use in commercial banking credit and credit-related decision-making. This paper proposes a three-way decision method based on prospect theory and evidence theory for predicting credit risk. The first problem in this study is determining the optimal classification boundary, and the second is effectively predicting the sample default status within an uncertain boundary. To address the limitation of the SVDD model, which is that it does not consider the aggregation degree, a new sample-weighted support vector data description (SW-SVDD) model is constructed by ranking samples according to their relative membership degree. The classification boundaries of the default and nondefault samples are determined according to the maximum prediction accuracy of the SW-SVDD model. The samples are divided into definite boundary nondefault, definite boundary default, and uncertain boundary samples. The default status of samples falling into definite boundaries is predicted by the SW-SVDD model. The three-way decision method combining prospect theory and evidence theory predicts the default status of samples falling into uncertain boundaries. This paper also proposes a new interpretability method based on the default probability, nondefault probability, and optimal threshold point obtained by the three-way decision model. The empirical results show that the proposed three-way decision method is a model that balances accuracy and interpretability. It has a higher classification performance than traditional models and can reveal the key features that lead to each customer's default. The proposed three-way decision method can enhance the accuracy and reliability of risk assessments, enabling financial institutions to make more informed lending decisions and more effectively manage credit portfolios.

A Comparative Study on Various ANN Optimization Algorithms for Magnetorheological Elastomer Carbonyl Iron Particle Concentration Estimation

Estimation particle composition such as particle shape, size, and concentration are crucial prior to the fabrication process of magnetorheological elastomer (MRE) to avoid process repetition due to inaccurate formulation. Currently, most of MRE prediction model were purposely used to predict the rheological properties such as shear stress and dynamic modulus, known as forward model. Nonetheless, very few studies have been reported to be capable able of predicting particle composition particularly in MR materials, which known as inverse model. Therefore, this paper proposed a carbonyl iron particle (CIP) concentration based MRE prediction model using neural network algorithm. Neural network-based machine learning model is more approachable compared to conventional mathematical modelling approach due to easily identify trends and pattern while handling multi-variety data. Various optimization algorithms have been employed such as Adam, RMSprop, SGD, AdaGrad, and Nadam throughout the modelling process. As the results, given shear strain amplitude, magnetic flux density, storage modulus, and loss factor as model input, SGD gave the maximum prediction accuracy with 0.95 and 3.038 MPa of R2 and RMSE, respectively. Hence, this model can be the basis to the MRE material and devices development particularly as the tool to reduce costing and time consuming.

Optimizing photovoltaic thermal solar systems efficiency through advanced artificial intelligence driven thermal management techniques

Photovoltaic Thermal (PV/T) solar systems have the capacity to convert solar radiation into both electrical and thermal energy. Solar cells convert solar radiation into electricity, whereas solar thermal collectors gather excess energy and use it to heat cooling fluids and extract leftover heat from photovoltaic cells. Optimizing the design of the PV/T cooling system is critical for increasing overall system efficiency. This research describes a revolutionary AI approach for determining the ideal design for PV/T cooling systems. The suggested system has three major stages: data collection (DC), data preprocessing (DP), and prediction (PS). In the DC step, the collected data is preprocessed before being input into the suggested prediction model. The PS uses the Optimized Deep Neural Network (ODNN) to forecast water, air, and electrical efficiency. To improve the model's effectiveness, a new met-heuristic method, the Sand Cat Swarm Optimization (SCSO), is used in conjunction with the Deep Neural Network (DNN). The SCSO approach improves the DNN's hyperparameters to maximize prediction accuracy. The results show that ODNN outperforms in terms of Mean Absolute Error (MAE), Mean Squared Error (MSE), and R-squared (R2), with values of approximately 0.0075, 0.00251, and 0.9972, respectively, demonstrating the effectiveness of the proposed AI-driven approach in optimizing PV/T cooling system design. In addition to predicting water, air, and electricity efficiency, the proposed AI-driven approach in this study shows the capacity to optimize the operational efficiency of a system. When combining SCSO with DNN, it shows its effectiveness in dynamically adjusting hyperparameters, contributing to the robustness of the model across many scenarios. Furthermore, the study focuses on the optimized PV/T cooling system's possible real-world applications, emphasizing its role in supporting sustainable energy solutions and solving the developing issues of solar energy consumption. The results confirm the suggested AI system's versatility and reliability, positioning it as a promising tool for improving solar energy technologies.