Regression Neural Network Research Articles

Prediction of Total Soluble Solids Content Using Tomato Characteristics: Comparison Artificial Neural Network vs. Multiple Linear Regression

Tomatoes are among the world’s most significant vegetables, both in terms of production and consumption. Harvesting takes place in tomato production when the important quality attribute of total soluble solids content reaches its maximum possible level. Tomato total soluble solids content (TSS) is among the most crucial attribute parameters for assessing tomato quality and for tomato commercialization. Determination of total soluble solids content by conventional measurement methods is both destructive and time-consuming. Therefore, the tomato processing industry needs a rapid identification method to measure total soluble solids content (TSS). In this study, we aimed to estimate how much soluble solids there are in beef tomato fruit by Artificial Neural Networks (ANN) and Multiple Linear Regression (MLR) methods. The models were assessed using the Coefficient of Determination (R2), Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and Mean Absolute Percentage Error (MAPE) metrics. The training data set results of the MLR model established to estimate the amount of brix in tomato fruit, calculated as MAE: 0.2349, RMSE: 0.3048, R2: 0.8441, and MAPE: 5.5368, while, according to the ANN model, MAE: 0.0250, RMSE: 0.031, R2: 0.9982 and MAPE: 0.5814. According to the metric outcomes, the ANN-based model performed better in both the training and testing parts.

PREDICTING RISK OF EMPLOYEE INJURY IN A PEDIATRIC HOSPITAL

PURPOSEHealthcare has one of the highest rates of non-fatal occupational injury as compared to other industries. Yet, evaluation of risk determinants and prediction algorithms in hospital settings remains limited. METHODSThis study examines risk factors for employee injuries in a large pediatric hospital and evaluates prediction algorithms using hospital surveillance data, incident reports, and work unit measures such as patient density and employee workload. We employed multiple statistical models and machine learning tools, including logistic regression (LR), random forest (RF), penalized logistic regression (PLR), Naïve Bayes, neural network (NN), XGBoost (XG), and mixed-effects logistic regression (GLMER) to predict employee injury risk for specific time periods. We used cross-validation and receiver-operator characteristic (ROC) curve analyses to assess model performance. RESULTSGLMER, LR, and PLR were superior to other models, with higher AUC values (∼0.76), indicating good discrimination ability, though accuracy and specificity varied across models. RF showed high accuracy and specificity and comparable AUC with the top performing models. Further analyses using GLMER revealed variability in employee injury risk across months, days, and hospital units, identifying peaks on Tuesdays and Saturdays and in April and July, with lows in March and June. CONCLUSIONOur findings highlight the significance of monitoring specific risk factors within pediatric hospital settings and pairing them with appropriate predictive algorithms to effectively predict and mitigate employee injuries. These insights indicate that continuous monitoring may help enhance employee safety. Future work should evaluate additional predictors that may be obtained from individual hospital units, which may inform targeted prevention strategies.

Performance Assessment of One-Part Self-Compacted Geopolymer Concrete Containing Recycled Concrete Aggregate: A Critical Comparison Using Artificial Neural Network (ANN) and Linear Regression Models

Geopolymer concrete, a cement-free concrete with recycled concrete aggregate (RCA), offers an eco-friendly solution for reducing carbon emissions from cement production and reusing a significant amount of old concrete from construction and demolition waste. This research on self-compacted, ambient-cured, and low-carbon concrete demonstrates the superior performance of one-part geopolymer concrete made from recycled materials. It is achieved by optimally replacing treated RCA with a unique method that involves coating the recycled aggregates with a one-part geopolymer slurry composed of fly ash, micro fly ash, slag, and anhydrous sodium metasilicate. The research presented in this paper introduces predictive models to assist researchers in optimising concrete mix designs based on RCA rates and treatment methods, including the incorporation of coated recycled concrete aggregates and basalt fibres. This study addresses the knowledge gap regarding geopolymer concrete based on recycled aggregate, various RCA rates, and novel RCA treatments. The novelty of the paper also lies in presenting the effectiveness of Artificial Neural Network (ANN) models in accurately predicting the compressive strength, splitting tensile strength, and modulus of elasticity for self-compacting geopolymer concrete with various rates of RCA replacement. This addresses a knowledge gap in existing research on ANN models for the prediction of geopolymer concrete properties based on RCA rate and treatment. The ANN models developed in this research predict results that are more comparable to experimental outcomes, showcasing superior accuracy compared to linear regression models.

Statistical and machine learning models for location-specific crop yield prediction using weather indices.

Crop yield prediction gains growing importance for all stakeholders in agriculture. Since the growth and development of crops are fully connected with many weather factors, it is inevitable to incorporate meteorological information into yield prediction mechanism. The changes in climate-yield relationship are more pronounced at a local level than across relatively large regions. Hence, district or sub-region-level modeling may be an appropriate approach. To obtain a location- and crop-specific model, different models with different functional forms have to be explored. This systematic review aims to discuss research papers related to statistical and machine-learning models commonly used to predict crop yield using weather factors. It was found that Artificial Neural Network (ANN) and Multiple Linear Regression were the most applied models. Support Vector Regression (SVR) model has a high success ratio as it performed well in most of the cases. The optimization options in ANN and SVR models allow us to tune models to specific patterns of association between weather conditions of a location and crop yield. ANN model can be trained using different activation functions with optimized learning rate and number of hidden layer neurons. Similarly, the SVR model can be trained with different kernel functions and various combinations of hyperparameters. Penalized regression models namely, LASSO and Elastic Net are better alternatives to simple linear regression. The nonlinear machine learning models namely, SVR and ANN were found to perform better in most of the cases which indicates there exists a nonlinear complex association between crop yield and weather factors.

Optimization of aerodynamic drag reduction for truck trailer model via machine learning

From advanced fairing designs and spoilers to novel airflow management systems, the quest for drag reduction (DR) in truck trailers is critical for cleaner and more sustainable transport. In this study, a particle swarm optimization (PSO) integrated machine learning approach is developed to optimize the DR performance of a bare truck trailer (BTT) with and without half airfoils and roof fairings. The study aims to reach the optimal truck trailer design with fewer experiments by searching the entire solution space with machine learning methods thanks to the developed approach. For this purpose, the experimental setup was initially designed and the experimental results were obtained. The experimental inputs are Airfoil thickness, Roof Fairing (RF) positions, and Reynolds number, while the experimental output is the drag coefficient. These inputs and outputs were arranged via data preprocessing and then modeled using machine learning methods. Modeling was performed with Artificial Neural Network (ANN) and multiple regression models. In addition to the ANN model setup with appropriate topology, linear and non-linear multiple regression models were established. Comparisons were performed using statistical metrics and it was determined that ANN was the superior model. Using the ANN’s prediction model, the drag coefficients were derived from the whole parameter values of both airfoil thickness and RF positions in the solution space for all Reynolds number and an optimization process was performed by PSO. As a result of the analysis, optimal parameter values were determined as 0.10 and 5.0 mm for the airfoil thickness and RF position. As a result of the experimental validation, it is seen that the BTT_NACA 0010_RF-0.5 model, designed with the optimum parameter values, demonstrated a DR performance of 34.7 %, which is the highest value compared to others. Thus, the heuristic process was experimentally validated and optimized by saving time and cost.

Evolution of structural safety for tunnel arch crown linings under the coupled influence of void and insufficient thickness

The existence of voids behind tunnel linings constitutes a recognized inherent defect in tunnel engineering. Void and insufficient thickness can significantly compromise structural safety, especially at the tunnel arch crown. In this study, we employed similar model tests and numerical analysis methods to examine how the coupling of voids and insufficient thickness affects the structural performance of arch crown linings. Key findings reveal that the coupling of arch crown voids and insufficient thickness causes notable shifts in displacement patterns and internal force distribution. The displacement pattern of the arch shifts from the inner side to the outer side of the tunnel, resulting in the concrete stress state on the inner side of the arch crown changing to compression. Spatial distribution characteristics of voids and insufficient thickness defects were analyzed to assess their impact on structural degradation, with quantitative analysis using the safety factor index. Geometric parameters of voids and insufficient thickness (longitudinal length, thickness, circumferential range) were chosen based on analysis under various conditions to predict the structural safety factor using a Wide Neural Network regression model with 98 % accuracy. This method offers a new approach for rapidly predicting the safety status of lining structures. These findings are crucial for comprehending the effects of arch crown voids and insufficient thickness defects and for devising effective strategies to address tunnel lining defects.

Construction and evaluation of a predictive model for the types of sleep respiratory events in patients with OSA based on hypoxic parameters.

To explore the differences and associations of hypoxic parameters among distinct types of respiratory events in patients with obstructive sleep apnea (OSA) and to construct prediction models for the types of respiratory events based on hypoxic parameters. A retrospective analysis was conducted on a cohort of 67 patients with polysomnography (PSG). All overnight recorded respiratory events with pulse oxygen saturation (SpO2) desaturation were categorized into four categories: hypopnea (Hyp, 3409 events), obstructive apnea (OA, 5561 events), central apnea (CA, 1110 events) and mixed apnea (MA, 1372 events). All event recordings were exported separately from the PSG software as comma-separated variable (.csv) files, which were imported into custom-built MATLAB software for analysis. Based on 13 hypoxic parameters, artificial neural network (ANN) and binary logistic regression (BLR) were separately used for construction of Hyp, OA, CA and MA models. Receiver operating characteristic (ROC) curves were employed to compare the various predictive indicators of the two models for different respiratory event types, respectively. Both ANN and BLR models suggested that 13 hypoxic parameters significantly influenced the classification of respiratory event types; The area under the ROC curves of the ANN models surpassed those of traditional BLR models respiratory event types. The ANN models constructed based on the 13 hypoxic parameters exhibited superior predictive capabilities for distinct types of respiratory events, providing a feasible new tool for automatic identification of respiratory event types using sleep SpO2.

Regression convolutional neural network models implicate peripheral immune regulatory variants in the predisposition to Alzheimer's disease.

Alzheimer's disease (AD) involves aggregation of amyloid β and tau, neuron loss, cognitive decline, and neuroinflammatory responses. Both resident microglia and peripheral immune cells have been associated with the immune component of AD. However, the relative contribution of resident and peripheral immune cell types to AD predisposition has not been thoroughly explored due to their similarity in gene expression and function. To study the effects of AD-associated variants on cis-regulatory elements, we train convolutional neural network (CNN) regression models that link genome sequence to cell type-specific levels of open chromatin, a proxy for regulatory element activity. We then use in silico mutagenesis of regulatory sequences to predict the relative impact of candidate variants across these cell types. We develop and apply criteria for evaluating our models and refine our models using massively parallel reporter assay (MPRA) data. Our models identify multiple AD-associated variants with a greater predicted impact in peripheral cells relative to microglia or neurons. Our results support their use as models to study the effects of AD-associated variants and even suggest that peripheral immune cells themselves may mediate a component of AD predisposition. We make our library of CNN models and predictions available as a resource for the community to study immune and neurological disorders.

Urban pluvial flood susceptibility mapping based on a novel explainable machine learning model with synchronous enhancement of fitting capability and explainability

Urban pluvial flood susceptibility mapping aims to identify flood-prone areas in cities. Machine learning methods are increasingly used, but typically face a trade-off between accurate identification of flood-susceptible areas and development of effective local mitigation strategies. To synchronously enhance fitting capability and explainability, this study introduced a novel machine learning model, Residual-Logistic Regression (RES-LR). RES-LR is developed by integrating Artificial Neural Network (ANN) and Logistic Regression (LR). The model adopts ANN to form its explainability module, which increases model complexity to ensuring the model’s fitting ability. Meanwhile, the explainability module was incorporated into the LR structure through residual connections to provide parametric expressions for local feature contributions, enabling the identification of key flood conditioning factor by site. The model was applied to the Beijing Metropolitan Area, achieving an accuracy rate exceeding 80% in identifying flooded sites. Based on its local explainability, the model unveiled the diverse flood formation mechanism in the study area and the complex impact of drainage-related factors. These results were informative for formulating site-specific flood mitigation measure. This study represents a significant advancement in the application of explainable machine learning for systematic flood management.

Prediction of 90day readmission in heart failure with preserved ejection fraction by interpretable machine learning.

Certain critical risk factors of heart failure with preserved ejection fraction (HFpEF) patients were significantly different from those of heart failure with reduced ejection fraction (HFrEF) patients, resulting in the limitations of existing predictive models in real-world situations. This study aimed to develop a machine learning model for predicting 90day readmission for HFpEF patients. Data were extracted from electronic health records from 1 August 2020 to 1 August 2021 and follow-up records of patients with HFpEF within 3months after discharge. Feature extraction was performed by univariate analysis combined with the least absolute shrinkage and selection operator (LASSO) algorithms. Machine learning models like eXtreme Gradient Boosting (XGBoost), random forest, neural network and logistic regression were adopted to construct models. The discrimination and calibration of each model were compared, and the Shapley Additive exPlanations (SHAP) method was used to explore the interpretability of the model. The cohort included 746 patients, of whom 103 (13.8%) were readmitted within 90days. XGBoost owned the best performance [area under the curve (AUC)=0.896, precision-recall area under the curve (PR-AUC)=0.868, sensitivity=0.817, specificity=0.837, balanced accuracy=0.827]. The Kolmogorov-Smirnov (KS) statistic was 0.694 at 0.468 in the XGBoost model. SHAP identified the top 12 risk features, including activities of daily living (ADL), left atrial dimension (LAD), left ventricular end-diastolic diameter (LVDD), shortness, nitrates, length of stay, nutritional risk, fall risk, accompanied by other symptoms, educational level, anticoagulants and edema. Our model could help medical agencies achieve the early identification of 90day readmission risk in HFpEF patients and reveal risk factors that provide valuable insights for treatments.

Advancing soil property prediction with encoder-decoder structures integrating traditional deep learning methods in Vis-NIR spectroscopy

The technology for estimating soil properties using visible and near-infrared spectroscopy has been maturing, with corresponding advances and breakthroughs in deep learning models. In this study, based on the large soil spectral library LUCAS, we explore the potential of encoder-decoder structures to improve convolutional neural network regression predictions. By introducing an encoder-decoder structure into the feature channels of a six-layer CNN model (TRNN model), we significantly enhanced the performance of shallow CNN models and successfully carried out regression predictions for seven soil properties. We employed IntegratedGradients, DeepLift, GradientShap, and DeepLiftShap methods to interpret the output of the TRNN model. Our TRNN model, built on raw spectra, demonstrated high accuracy in predicting multiple soil properties, outperforming residual architectures, LSTMs, various CNN architectures, and other traditional machine learning methods proposed in previous studies. We also investigated the impact of multi-task output structures (TRNN 1-M and TRNN M−M) and single-task output structures (TRNN 1-1) on model performance. For the TRNN model with an encoder-decoder structure, multi-task output structures resulted in a reduction in performance. The TRNN showed outstanding results in regression analysis of the seven soil properties selected in this study (cation exchange capacity, organic carbon content, calcium carbonate content, pH, clay content, silt content, and sand content), with R2 values exceeding 0.93 for all seven properties. Different soil characteristics correspond to different wavelengths, with multiple characteristic peaks commonly observed. This research convincingly demonstrates the enormous potential of combining large model architectures with traditional deep learning approaches for predicting soil properties, which could significantly advance precision agriculture.

Machine learning based methods for ratemaking health care insurance

In insurance, proposing an accurate premium that is adjusted to the insured risk profile allows companies to better manage their portfolios and to be more competitive. Machine learning methods have recently been adopted for various improvements in insurance ratemaking, especially in the automobile industry. These models are specifically used to mine potential data information and to build a predictive model for a variable of interest using explanatory variables. In this paper, we aim to provide a pricing method for ratemaking individual healthcare insurance contracts using machine learning algorithms that are applied to a Tunisian healthcare insurance portfolio. We start with a simple Classification and Regression Tree, and we work toward more advanced methods that are Random Forest, Extreme gradient boosting, Support Vector Regression, and Artificial Neural network regression model. The predictive performance of these non-parametric methods is compared with the standard generalized linear model. Our results showed the applicability of machine learning in the healthcare insurance market and that the XGBoost algorithm outperforms the predictive capacity of the classical generalized linear model.

Development of an efficient design optimization strategy for thick-walled cylinders treated with combinations of autofrettage, shrink-fit and wire-winding processes

Shrink-fit, wire-winding, and autofrettage processes are commonly utilized to enhance fatigue strength and durability of thick-walled cylinders across various mechanical applications. In this study, a novel practical design optimization methodology has been developed to determine the optimal configuration of a thick-walled cylinder, incorporating different combinations of shrink-fit, wire-winding, and autofrettage techniques. The objective is to identify the optimal layer thickness, shrink-fit interference, conventional autofrettage pressure, and reverse autofrettage pressure, if applicable, to maximize the compressive residual stress and minimize the tensile residual stress, thereby extending fatigue lifetime of the cylinder. First, different configurations of thick-walled cylinders, subjected to various combinations of reinforcement processes, are identified. A dataset of residual hoop stress profiles through the cylinder thickness is subsequently generated for these configurations based on the same manufacturing process. Neural network regression is effectively utilized to construct a single fitting function for the residual hoop stress profiles. A parametric study is performed to determine the optimal training functions, activation functions, and hyperparameters, achieving a remarkable agreement with the dataset, indicated by a coefficient of determination of over 0.97. A combination of Genetic Algorithm and Sequential Quadratic Programming algorithms is utilized to determine the accurate optimal values. Fatigue life analysis is subsequently conducted to estimate the fatigue lifetime of the optimal configuration. Results suggest that the optimal configuration, involving conventional autofrettage of the inner layer followed by shrink-fitting with a virgin layer and wire-winding the entire assembly, achieves a maximum fatigue life of 88 × 10⁶ cycles under cyclic pressure load of 300 MPa.

Study on sporty exhaust sound of economical vehicle under acceleration

Exhaust sound quality is an important part of vehicle performance. In this paper, the sporty exhaust sound quality of an economical vehicle equipped with a 4-cylinder and 4-stroke engine is evaluated, analyzed, and improved under acceleration. Firstly, a sporty feeling evaluation method with engine speed divided is proposed, and the influence of exhaust sound order components on sporty exhaust sound is analyzed. The results show that while the A-weighted sound pressure level (ASPL) of Order 2 is lower and the ASPLs of Orders 4 and 6 are higher, the exhaust sound is sportier. Then, a hybrid predicted model of vehicle sporty exhaust sound under acceleration is established based on convolutional neural network (CNN) and support vector regression (SVR) algorithm. The relative errors between the predicted results of CNN-SVR hybrid model and the subjective evaluation results are limited within 2%, which indicates that the CNN-SVR hybrid prediction model achieves a high accuracy in assessing the sporty feeling of exhaust sound. Finally, considering the frequency ranges corresponding with the above order components under the practical accelerating condition, a strategy is proposed to enhance the sporty feeling of exhaust sound by reducing the sound energy within 100 Hz and increasing the sound energy within 100–450 Hz. Based on this strategy, a muffler with different structure is selected and installed on the economical vehicle, and the sporty feeling of exhaust sound is 0.63 points higher than before.

Optimizing stock market volatility predictions based on the SMVF-ANP approach

The stock market is considered one of the most complicated financial systems, comprising several components or inventories, whose prices vary substantially over time. Bursaries include revealing market tendencies over time. All investors in the stock market aim to maximize profits and reduce the associated risks. Stock currency forecasts are a significant financial concern that is being handled even more closely. In recent years, various neural network and hybrid models have surpassed classic linear and non-linear techniques to produce reliable prediction outcomes. This study investigates the efficacy of dynamic and effective stock market forecasting using neural network models. This study examines market transmission mechanisms and assesses the predicted links between multiple financial and economic factors. A stock market volatility and artificial network prediction (SMVF–ANP) approach is presented. The models analyzed included a multi-layer perceptron (MLP), dynamic artificial neural network, and generalized autoregressive conditional heteroscedasticity to extract additional input variables. The results reveal that the trade strategies led by the classification models yield superior risk-adjusted returns compared to the buy-and-hold approach and those led by the neural network and linear regression model-level estimate predictions. The numerical results show that the proposed SMVF-ANP method achieved a high performance ratio of 94.1%, an enhanced prediction ratio of 98.4%, a high stock market volatility rate of 96.7%, a reduced mean square percentage error ratio of 16.3%, a probability rate of 32.7%, an increased efficiency rate of 96.9%, and an accuracy ratio of 97.2% compared to other methods.

Confinement strength prediction of corroded rectangular concrete columns using BP neural networks and support vector regression

The application of machine learning in predicting the mechanical performance of reinforced concrete columns subjected to reinforcement corrosion was investigated in this study. Traditional models, often relying on empirical formulas or simplified assumptions, are known to struggle in capturing all variables and relationships in complex problems, leading to inadequate prediction accuracy. Through theoretical analysis, the confinement effect of stirrups on core concrete and its performance deterioration caused by reinforcement corrosion were determined. To improve the prediction accuracy and generalization ability, the error Back Propagation Neural Network (BPNN) and Support Vector Regression (SVR) models were employed. Parameter effects in the BPNN model were interpreted using the SHapley Additive exPlanations (SHAP) and compared with traditional parameter sensitivity analyses. The investigation indicates that the machine learning models have a high degree of fit and low error levels, particularly within the training set. After analyzing using the SHAP approach, it is found that the corrosion ratio of the stirrup and longitudinal reinforcement have a significant effect on the confinement strength of the concrete, whereas the effects of the stirrup configuration and size effect are negligible, which is different from conventional parametric sensitivity analysis. As to the stability analysis of model predictions, the SVR model shows lower volatility and higher prediction stability compared with the BPNN model, despite the latter occasionally providing superior generalization ability. This research underscores the significant advantages offered by machine learning models in addressing longstanding challenges associated with strength prediction, and thus presents new perspectives for future study.

Prediction Of Student Achievement Using Artificial Neural Network And Support Vector Regression At SMK TELKOM Lampung

The analysis of student performance is crucial in vocational schools because it helps identify the challenges students face in preparing themselves for the workforce. By integrating data mining techniques such as Artificial Neural Networks (ANN), educators can enhance their understanding of factors that improve student learning outcomes. An artificial neural network (ANN) is composed of interconnected artificial neurons that can learn from input data and make complex predictions, including academic achievements. The structure and function of the human brain inspire ANN. This study compares the effec- tiveness of the artificial neural network (ANN) method with other methodologies, such as support vector regression (SVR), to predict student achievement at SMK Telkom Lampung. Primary data collected from SMK Telkom Lampung includes 4939 examples with 550 cases, 26 features, and 4 meta-attributes. Performance evaluation involves metrics such as Mean Squared Error (MSE), Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), and coefficient of determination (R2). The coefficient of determination (R2) value of the Neural Network at 0.001 is higher than the R2 value of SVR, which reaches -0.036. Research find- ings indicate that the Artificial Neural Network model slightly outperforms the Support Vector Regression model, with lower prediction error rates and better ability to explain data variability.

Prediction of flash flood peak discharge in hilly areas with ungauged basins based on machine learning

ABSTRACT Peak discharge is an essential element of hydrological forecasting. Due to rapid outbreaks of flash floods in hilly areas and the lack of measured data, the fast and accurate estimation of peak discharge is crucial for flash flood hazard management. Three machine learning algorithms were applied to estimate peak discharge; this estimation was compared with the results of hydrological–hydraulic models, and the results were verified with measured watershed data. In this paper, 10 hydrological and geomorphological parameters were selected to predict the flood peak discharge in 103 watersheds in Taiyi Mountain North District. The results show that the particle swarm optimization backpropagation (PSO-BP) neural network model outperforms the BP neural network and random forest regression in prediction performance. PSO-BP has a lower mean absolute error (2.51%), root mean square error (3.74%), and mean absolute percentage error (2.74%) than the other models, which indicates that PSO-BP has high prediction accuracy. Importance analysis revealed that rainfall, early impact rainfall, catchment area, and rain intensity are the key input parameters of PSO-BP. The proposed method was confirmed to be a fast and relatively accurate algorithm for estimating the peak discharge of flash floods in ungauged basins.

Enhancing Fermentation Process Monitoring through Data-Driven Modeling and Synthetic Time Series Generation.

Soft sensors based on deep learning regression models are promising approaches to predict real-time fermentation process quality measurements. However, experimental datasets are generally sparse and may contain outliers or corrupted data. This leads to insufficient model prediction performance. Therefore, datasets with a fully distributed solution space are required that enable effective exploration during model training. In this study, the robustness and predictive capability of the underlying model of a soft sensor was improved by generating synthetic datasets for training. The monitoring of intensified ethanol fermentation is used as a case study. Variational autoencoders were employed to create synthetic datasets, which were then combined with original datasets (experimental) to train neural network regression models. These models were tested on original versus augmented datasets to assess prediction improvements. Using the augmented datasets, the soft sensor predictive capability improved by 34%, and variability was reduced by 82%, based on R2 scores. The proposed method offers significant time and cost savings for dataset generation for the deep learning modeling of ethanol fermentation and can be easily adapted to other fermentation processes. This work contributes to the advancement of soft sensor technology, providing practical solutions for enhancing reliability and robustness in large-scale production.

Mathematical optimization of multilinear and artificial neural network regressions for mineral composition of different tea types infusions

The objective of this study was to investigate the change in mineral composition depending on tea variety, tea concentration, and steeping time. Four different tea varieties, black Ceylon (BC), black Turkish (BT), green Ceylon (GC), and green Turkish (GT), were used to produce teas at concentrations of 1, 2, and 3%, respectively. These teas were produced using 7 different steeping times: 2, 5, 10, 20, 30, 45, and 60 min. It was also aimed to optimize the regression equations utilizing these factors to identify parameters conducive to maximizing Zn, K, Cu, Mg, Ca, Na, and Fe levels; minimizing Al content, and maintaining Mn level at 5.3 mg/L. The optimal conditions for achieving a Mn content of 5.3 mg/L in black Turkish tea entailed steeping at a concentration of 1.94% for 11.4 min. Variations in K and Mg levels across teas were inconsistent with those observed for other minerals, whereas variations in Al, Cu, Fe, Mn, Na, and Zn levels exhibited a close relationship. Overall, mineral levels in tea can be predicted through regression analysis, and by mathematically optimizing the resultant equations, the requisite conditions for tea production can be determined to achieve maximum, minimum, or target mineral values.