Grid Search Research Articles

Intelligent cardiovascular disease diagnosis using deep learning enhanced neural network with ant colony optimization

To identify patterns in big medical datasets and use Deep Learning and Machine Learning (ML) to reliably diagnose Cardio Vascular Disease (CVD), researchers are currently delving deeply into these fields. Training on large datasets and producing highly accurate validation results is exceedingly difficult. Furthermore, early and precise diagnosis is necessary due to the increased global prevalence of cardiovascular disease (CVD). However, the increasing complexity of healthcare datasets makes it challenging to detect feature connections and produce precise predictions. To address these issues, the Intelligent Cardiovascular Disease Diagnosis based on Ant Colony Optimisation with Enhanced Deep Learning (ICVD-ACOEDL) model was developed. This model employs feature selection (FS) and hyperparameter optimization to diagnose CVD. Applying a min–max scaler, medical data is first consistently prepared. The key feature that sets ICVD-ACOEDL apart is the use of Ant Colony Optimisation (ACO) to select an optimal feature subset, which in turn helps to upgrade the performance of the ensuring deep learning enhanced neural network (DLENN) classifier. The model reforms the hyperparameters of DLENN for CVD classification using Bayesian optimization. Comprehensive evaluations on benchmark medical datasets show that ICVD-ACOEDL exceeds existing techniques, indicating that it could have a significant impact on CVD diagnosis. The model furnishes a workable way to increase CVD classification efficiency and accuracy in real-world medical situations by incorporating ACO for feature selection, min–max scaling for data pre-processing, and Bayesian optimization for hyperparameter tweaking.

Machine learning and deep learning models based grid search cross validation for short-term solar irradiance forecasting

In late 2023, the United Nations conference on climate change (COP28), which was held in Dubai, encouraged a quick move from fossil fuels to renewable energy. Solar energy is one of the most promising forms of energy that is both sustainable and renewable. Generally, photovoltaic systems transform solar irradiance into electricity. Unfortunately, instability and intermittency in solar radiation can lead to interruptions in electricity production. The accurate forecasting of solar irradiance guarantees sustainable power production even when solar irradiance is not present. Batteries can store solar energy to be used during periods of solar absence. Additionally, deterministic models take into account the specification of technical PV systems and may be not accurate for low solar irradiance. This paper presents a comparative study for the most common Deep Learning (DL) and Machine Learning (ML) algorithms employed for short-term solar irradiance forecasting. The dataset was gathered in Islamabad during a five-year period, from 2015 to 2019, at hourly intervals with accurate meteorological sensors. Furthermore, the Grid Search Cross Validation (GSCV) with five folds is introduced to ML and DL models for optimizing the hyperparameters of these models. Several performance metrics are used to assess the algorithms, such as the Adjusted R2 score, Normalized Root Mean Square Error (NRMSE), Mean Absolute Deviation (MAD), Mean Absolute Error (MAE) and Mean Square Error (MSE). The statistical analysis shows that CNN-LSTM outperforms its counterparts of nine well-known DL models with Adjusted R2 score value of 0.984. For ML algorithms, gradient boosting regression is an effective forecasting method with Adjusted R2 score value of 0.962, beating its rivals of six ML models. Furthermore, SHAP and LIME are examples of explainable Artificial Intelligence (XAI) utilized for understanding the reasons behind the obtained results.

Event locations: speeding up gridsearches using quadratic interpolation

SUMMARY The grid search method is a common approach to estimate the three spatial coordinates of event hypocentres. However, locating events in large search spaces with small grid spacings is computationally prohibitive. This study accelerates the grid searches over large search spaces using a quadratic interpolation technique. We start with the coarse-grid-estimated location, where we have the minimum value of the difference in the traveltimes between S and P waves summed over all receivers. Then, we select the neighbouring grid points and build a 3D quadratic function. The unknown coefficients of the 3D quadratic function are computed by solving a system of linear equations. After that, we interpolate the location by solving partial derivatives of the quadratic function. The quadratic interpolation technique performs well on both synthetic and real microseismic data examples, typically leading to similar event locations as those obtained using 10 times smaller grid spacings in all three directions, at a minor additional computational expense, and without the need to generate traveltimes at new spatial positions.

Relating rainfall retrieval parameters to network and environmental features to improve rainfall estimates from commercial microwave links in the tropics

Abstract Potentially the greatest benefit of Commercial Microwave Links (CMLs) as opportunistic rainfall sensors lies in regions that lack dedicated rainfall sensors, most notably low- and middle income countries. However, current CML rainfall retrieval algorithms are predominantly tuned and applied to (European) CML networks in temperate or Mediterranean climates. This study investigates whether local quantitative precipitation estimates from CMLs in a tropical region, specifically Sri Lanka, can be improved by optimizing two dominant parameters in the rainfall retrieval algorithm RAINLINK, namely the wet-antenna correction factor Aa and the relative contribution of minimum and maximum received signal levels α. Using a grid search, based on ten months of CML data from 22 link-gauge clusters consisting of 105 sub-links that lie within 1 km of a daily rain gauge, optimal values of Aa and α are first derived for the entire country and compared to the default RAINLINK values. Subsequently, the CMLs are grouped by link length, frequency, climate zone, and daily rainfall depth classes, and Aa and α are derived for each of these classes. Calibrating parameters on all clusters across the country only leads to minor improvements. The actual optimal Aa and α values depend on the performance metric favored. Calibrating on network properties, particularly short link length and high frequency classes, does significantly improve rainfall estimates. By relating the optimal Aa and α values to known network meta data, the results from this study are potentially applicable to other tropical CML networks that lack nearby reference rainfall data.

Mapping Glasgow Coma scale to AVPU scores among adults in the prehospital setting

ObjectiveTo develop a translation between the Glasgow Come Scale and the Alert-Verbal-Pain-Unresponsive (AVPU) scale among adults with out-of-hospital emergencies. MethodsWe performed a retrospective analysis of adults (≥18 years) from the 2022 National Emergency Medical Services (EMS) Information System with a ground scene encounter with a concurrently documented GCS and AVPU assessment. Using a training partition of 2.5 million encounters, we performed a grid search to identify all combinations of mutually exclusive cutpoints which divided the GCS into four segments. We identified the combination with the highest Kappa statistic and reported metrics of performance in this sample in the test partition. ResultsWe identified 16,321,299 encounters with a concurrent AVPU and GCS. Using the AVPU scale, 93.3 % were classified as Alert; 2.9 % as Verbal; 1.5 % as Pain; and 2.3 % as Unresponsive. Using a grid-based search, optimal cutpoints were identified when using a GCS of 14–15 for Alert, 10–13 for Verbal, 7–9 for Pain, and 3–6 for Unresponsive. Cohen's Kappa was 0.63 in the test partition, indicating substantial agreement. Intraclass F1 score varied across different alertness levels and were 0.97 for “Alert”, 0.43 for “Verbal”, 0.49 for “Pain”, and 0.83 for “Unresponsive”. Findings were similar in analyses performed by age group and by the presence or absence of trauma. ConclusionWe report an optimal crosswalk between the AVPU and GCS scales. Performance in the Verbal and Pain categories was lower than the Alert and Unresponsive categories. These findings may facilitate clinician handovers between EMS and non-EMS clinicians.

Bayesian Optimized Machine Learning Model for Automated Eye Disease Classification from Fundus Images

Eye diseases are defined as disorders or diseases that damage the tissue and related parts of the eyes. They appear in various types and can be either minor, meaning that they do not last long, or permanent blindness. Cataracts, glaucoma, and diabetic retinopathy are all eye illnesses that can cause vision loss if not discovered and treated early on. Automated classification of these diseases from fundus images can empower quicker diagnoses and interventions. Our research aims to create a robust model, BayeSVM500, for eye disease classification to enhance medical technology and improve patient outcomes. In this study, we develop models to classify images accurately. We start by preprocessing fundus images using contrast enhancement, normalization, and resizing. We then leverage several state-of-the-art deep convolutional neural network pre-trained models, including VGG16, VGG19, ResNet50, EfficientNet, and DenseNet, to extract deep features. To reduce feature dimensionality, we employ techniques such as principal component analysis, feature agglomeration, correlation analysis, variance thresholding, and feature importance rankings. Using these refined features, we train various traditional machine learning models as well as ensemble methods. Our best model, named BayeSVM500, is a Support Vector Machine classifier trained on EfficientNet features reduced to 500 dimensions via PCA, achieving 93.65 ± 1.05% accuracy. Bayesian hyperparameter optimization further improved performance to 95.33 ± 0.60%. Through comprehensive feature engineering and model optimization, we demonstrate highly accurate eye disease classification from fundus images, comparable to or superior to previous benchmarks.

A novel systematically optimized tabular neural network (TabNet) algorithm for predicting the tensile modulus of additively manufactured PLA parts

Assessing the elastic modulus of 3D-printed polylactic acid (PLA) components is essential for understanding their stiffness and load capacity, which are crucial for predicting product performance and durability. In this study, the predictive accuracy of a Tabular Neural Network (TabNet) algorithm for determining the elastic modulus of 3D-printed PLA components via fused deposition modeling (FDM) was investigated. Utilizing a comprehensive dataset of 128 literature-sourced data points, divided into 80 % for training and 20 % for validation, the study proposed a new Taguchi-based method for efficient hyperparameter optimization of the TabNet algorithm. This optimization revealed that a configuration of 8 decision blocks, 16 attention blocks, and 5 decision steps, along with the "Adam" optimizer, a gamma of 1, learning rate of 0.1, and lambda-sparse of 0.01, yielded the highest prediction accuracy for the elastic modulus of PLA parts. The performance of the optimized TabNet model was evaluated using R-squared (R²), Mean Absolute Error (MAE), Mean Squared Error (MSE), and Root Mean Squared Error (RMSE) measures. The findings highlighted an R² of 96.855 %, an MAE of 0.158, an MSE of 0.037, and an RMSE of 0.193 in the validation dataset, demonstrating substantial predictive reliability. To further test the model’s robustness, fourteen unseen data points were analyzed. The observed discrepancies between predicted and actual values were under 10 %, affirming the Taguchi-optimized TabNet algorithm’s effectiveness in forecasting the elastic modulus of FDM 3D-printed PLA components. This investigation provides a significant advancement in additive manufacturing, introducing a precise and reliable method for predicting the mechanical properties of 3D-printed materials.

Research on Critical Quality Feature Recognition and Quality Prediction Method of Machining Based on Information Entropy and XGBoost Hyperparameter Optimization

To address the problem of predicting machining quality for critical features in the manufacturing process of mechanical products, a method that combines information entropy and XGBoost (version 2.1.1) hyperparameter optimization is proposed. Initially, machining data of mechanical products are analyzed based on information entropy theory to identify critical quality characteristics. Subsequently, a quality prediction model for these critical features is established using the XGBoost machine learning framework. The model’s hyperparameters are then optimized through Bayesian optimization. This method is applied as a case study to a medium-speed marine diesel engine piston. After the critical quality characteristics in the machining process are identified, the machining quality of these vital characteristics is predicted, and the results are compared with those obtained from a machine learning model without hyperparameter optimization. The findings demonstrate that the proposed method effectively predicts the machining quality of mechanical products.

Automated machine learning with R: AutoML tools for beginners in clinical research.

Recently, interest in machine learning (ML) has increased as the application fields have expanded significantly. Although ML methods excel in many fields, establishing an ML pipeline requires considerable time and human resources. Automated ML (AutoML) tools offer a solution by automating repetitive tasks, such as data preprocessing, model selection, hyperparameter optimization, and prediction analysis. This review introduces the use of AutoML tools for general research, including clinical studies. In particular, it outlines a simple approach that is accessible to beginners using the R programming language (R Foundation for Statistical Computing). In addition, the practical code and output results for binary classification are provided to facilitate direct application by clinical researchers in future studies.

Classification of Unisba Students' Graduation Time using Support Vector Machine Optimized with Grid Search Algorithm

Support Vector Machine is a classification method that finds the optimal hyperplane to separate two data classes. SVM has much better generalization performance than other methods. However, SVM needs to improve in determining hyperparameter values. Therefore, parameter optimization is necessary to determine the optimal hyperparameter value. Grid search is one of the parameter optimization methods that can improve the quality of SVM models. This study aims to assess the level of accuracy in predicting student graduation times by using five features that affect it. This study shows that the resulting SVM model optimized with the Grid Search Algorithm is quite consistent and prevents overfitting. By utilizing the results of SVM modelling, UNISBA is expected to improve the quality of graduates. The risk of delays in graduation can be considered early by paying attention to the background and achievements of students

Distributed hybrid quantum-classical performance prediction for hyperparameter optimization

Hyperparameter optimization (HPO) of neural networks is a computationally expensive procedure, which requires a large number of different model configurations to be trained. To reduce such costs, this work presents a distributed, hybrid workflow, that runs the training of the neural networks on multiple graphics processing units (GPUs) on a classical supercomputer, while predicting the configurations’ performance with quantum-trained support vector regression (QT-SVR) on a quantum annealer (QA). The workflow is shown to run on up to 50 GPUs and a QA at the same time, completely automating the communication between the classical and the quantum systems. The approach is evaluated extensively on several benchmarking datasets from the computer vision (CV), high-energy physics (HEP), and natural language processing (NLP) domains. Empirical results show that resource costs for performing HPO can be reduced by up to 9% when using the hybrid workflow with performance prediction, compared to using a plain HPO algorithm without performance prediction. Additionally, the workflow obtains similar and in some cases even better accuracy of the final hyperparameter configuration, when combining multiple heuristically obtained predictions from the QA, compared to using just a single classically obtained prediction. The results highlight the potential of hybrid quantum-classical machine learning algorithms. The workflow code is made available open-source to foster adoption in the community.

Short‐term power prediction of distributed PV based on multi‐scale feature fusion with TPE‐CBiGRU‐SCA

AbstractTo address the challenge of insufficient comprehensive extraction and fusion of meteorological conditions, temporal features, and power periodic features in short‐term power prediction for distributed photovoltaic (PV) farms, a TPE‐CBiGRU‐SCA model based on multiscale feature fusion is proposed. First, multiscale feature fusion of meteorological features, temporal features, and hidden periodic features is performed in PV power to construct the model input features. Second, the relationships between PV power and its influencing factors are modelled from spatial and temporal scales using CNN and Bi‐GRU, respectively. The spatiotemporal features are then weighted and fused using the SCA attention mechanism. Finally, TPE‐based hyperparameter optimization is used to refine network parameters, achieving PV power prediction for a single field station. Validation with data from a PV field station shows that this method significantly enhances feature extraction comprehensiveness through multiscale fusion at both data and model layers. This improvement leads to a reduction in MAE and RMSE by 26.03% and 38.15%, respectively, and an increase in R2 to 96.22%, representing a 3.26% improvement over other models.

Optimized Machine Learning Classifiers for Symptom-Based Disease Screening

This work presents a disease detection classifier based on symptoms encoded by their severity. This model is presented as part of the solution to the saturation of the healthcare system, aiding in the initial screening stage. An open-source dataset is used, which undergoes pre-processing and serves as the data source to train and test various machine learning models, including SVM, RFs, KNN, and ANNs. A three-phase optimization process is developed to obtain the best classifier: first, the dataset is pre-processed; secondly, a grid search is performed with several hyperparameter variations to each classifier; and, finally, the best models obtained are subjected to additional filtering processes. The best-results model, selected based on the performance and the execution time, is a KNN with 2 neighbors, which achieves an accuracy and F1 score of over 98%. These results demonstrate the effectiveness and improvement of the evaluated models compared to previous studies, particularly in terms of accuracy. Although the ANN model has a longer execution time compared to KNN, it is retained in this work due to its potential to handle more complex datasets in a real clinical context.

A novel algorithm combined X-ray fluorescence and Neural Network (XRF-NN) for coal ash content prediction: Algorithm design and performance evaluation

This paper investigated a precise algorithm combining X-ray Fluorescence and Neural Network (XRF-NN) for predicting ash content. The 261 sets of XRF tests show that the 34 elements in the chosen Guqiao coal can be categorized as major, secondary and tiny elements, whose cumulative ratio were ~95%, 4-5% and <1%, respectively. Referring to the machine learning theory, the construction strategy of Element-Ash dataset was determined viz. value determination → standardization → division → optimization of generalization ability. Then, the hyperparameters optimization displays that XRF-NN model with 34 inputs and 1 output was suitable to predict coal ash content, where the activation function, loss function, and optimizer were ReLU, MAE, and Momentum, respectively. After iterative training, the new XRF-NN model provides the precise prediction of coal ash content with the absolute errors between -2.0% and 2%. Moreover, the prediction accuracy rose from 57.69% to 100%, as the expected relative error increased from 1% to 5%. Furthermore, the comparisons between different prediction methods reveal that the minimum MRE of 1.22% can be obtained by XRF-NN with total elements, which was only half of those given by the conventional Multiple Linear Regression and Partial Least Squares Regression. Besides, XRF-NN model presents the root mean squared error 0.797%, the mean absolute error 0.625% and the coefficient of determination 0.999, which were significantly superior to those calculated by Dual-Energy Gamma-ray Transmission, Ploy, RFR, XGBoost and DNN model. The results of this study suggest the excellent performance of the new XRF-NN model in predicting ash content.

Hyperparameter optimization of regional hydrological LSTMs by random search: A case study from Basque Country, Spain

This paper introduces a novel approach for hyperparameter optimization of long short-term memory networks (LSTMs) to achieve highly accurate hourly streamflow and water level predictions in the realm of regional rainfall-runoff modeling. Leveraging simultaneous systematic hyperparameter optimization of 10 distinct hyperparameters by Random Search, the study achieves high accuracy in terms of predictions across 40 humid flashy catchments in Basque Country, north of Spain. By carefully designing the search space and incorporating domain expertise, the approach quickly converges to optimal and highly accurate network configurations with both efficiency and efficacy. LSTMs ingested precipitation, temperature, and potential evapotranspiration as inputs to predict 2 targets of streamflow and water level, in an hourly timestep. On the test set, the optimized LSTM networks accurately predicted streamflow and water level with Nash-Sutcliffe (NSE) and Kling-Gupta (KGE) efficiencies as high as 0.97, in one of the catchments. Across all 40 studied catchments, the overall average NSE and KGE values for streamflow were 0.89 and 0.87, respectively; water level exhibited average NSE and KGE scores of 0.91 and 0.92.Moreover, statistical analysis reveals significant differences in the performance of the 2 distinct optimized network architectures in different hydrological catchments, underscoring the importance of deliberate network configuration selection post-random search. This selection process is vital for achieving higher performance in as many catchments as possible. The findings highlight opportunities for enhancing the “learning maturity” of regional hydrological deep learning LSTM networks. This research provides valuable insights for researchers and practitioners involved in optimizing regional hydrological deep learning models for a variety of applications and on new datasets.

Improved whale optimization algorithm towards precise state-of-charge estimation of lithium-ion batteries via optimizing LSTM

Accurate state-of-charge (SOC) estimation is a crucial part of the battery management system (BMS). However, conventional estimation methods are unable to capture the extremely complex dynamic characteristics of lithium-ion batteries. Besides, manually setting the optimal hyperparameters of models has many drawbacks. To address the problems, an (improved whale optimization algorithm) IWOA- (long short-term memory) LSTM model is proposed in this work. Utilizing the whale optimization algorithm (WOA) improved with four enhancement strategies (Gaussian chaotic mapping initialization, Nonlinear weight update, Lévy flight mechanism, and Elite opposition-based learning) to optimize the number of hidden layer nodes, the learning rate, and the number of iterations of LSTM model. It not only overcomes the shortcomings of artificially setting LSTM hyperparameters but also further boosts the learning ability of the IWOA-LSTM model, making the model more suitable for SOC estimation under different scenarios. The evaluation results show that the MAE of the proposed model for SOC estimation results is lower than 0.8 % under different temperatures and dynamic conditions. Compared with SOTA models, all MAE, RMSE, and MAPE of the proposed model substantially decline. Furthermore, the R2 of the estimation results using the LG dataset in Experiment Ⅲ is 98.65 %, suggesting the applicability of the proposed model to Li-ion batteries from various manufacturers. The experimental results demonstrate the proposed IWOA-LSTM model is suitable for accurate SOC estimation.

PSLBI-22 Identifying plasma metabolites influencing body weight in Salmonella challenged growing pigs: A machine learning approach

Abstract Body weight (BW) is an important component of pig productivity. Understanding the metabolic factors affecting growth rate is key to guiding nutritional strategies and bringing economic benefits to pig farms. Machine learning (ML) algorithms have been used to capture patterns in large datasets and predict animal production traits based on integrated information of performance data and blood records. However, utilizing ML models and metabolomic profiling to predict BW of pigs in small datasets with challenged animals has yet to be reported. Therefore, this study investigated use of plasma metabolites in predicting BW of pigs at the growing stage. This simulation used data from gilts (n = 40; BW = 25.6 ± 2.3 kg) raised over a 28-d trial. Plasma samples and BW were collected on d 0, 14, and 28 after 8 h of fasting. Ninety plasma metabolites were identified and quantified from an untargeted metabolomics assay. The development dataset included 30, and the validation dataset contained 10 unique animals with repeated measures in time. The metabolites were log-transformed, Pareto-scaled and missing values were imputed using the minimum nonzero value method. The BW prediction model was performed in R with the caret package. The optimal hyper-parameter configuration for the random forest and neural network algorithms was evaluated based on repeated cross-validation. The final model was the random forest composed of 400 trees with 45 randomly selected predictors. Model accuracy was estimated on repeated cross-validation with 10-folds and 5 repeats and compared with the validation set based on the coefficient of determination (R2), the root mean squared error (RMSE), and Pearson correlation (r). The metabolites with greater contribution with variable importance score above 10 to the model development were extracted. Aiming to understand differences in BW categories, the model response was visualized with an alluvial diagram for the 3 variables with the greatest metabolite importance, and BW was classified as high, medium, and low. The main metabolites driving BW were lysine, oleic acid, aminomalonic acid, myo-inositol, and linoleic acid. The RMSE after cross-validation was 7.22 kg of BW (r = 0.70), and R2 = 0.39. For pigs with a high BW, the predicted BW was explained by generally less plasmatic concentrations of lysine, oleic acid, and aminomalonic acid. Lysine is the main amino acid for muscle protein accretion, whereas oleic acid is for energy substrate. Aminomalonic acid is a metabolite resulting from glycine metabolism and is important in reversing oxidative damage caused by pathogens. Therefore, decreased plasma lysine, amino malonic acid, and oleic acid may indicate faster growth and lean deposition, leading to heavier BW in pigs. Using random forests helped to connect the plasma dynamics to BW and is a promising avenue to understand the main drivers in BW of group-housed pigs.

67 Precision identification and weight assessment of cattle using supervised machine learning on body surface keypoints

Abstract Computer vision systems (CVS) offer identification solutions for animals with distinct coat patterns, but are less effective for solid-colored herds. In addition, they can be used to measure critical phenotypes, such as body weight (BW). Providing both BW and identification for solid-colored animals can help farmers make decisions. This study aimed to 1) develop an automatic CVS capable of identifying using the Euclidean distance between keypoints located at specific anatomical landmarks (e.g., bony prominences), and 2) predict the BW using features extracted from these keypoints. The keypoint model was trained to identify seven keypoints per cow, located at the left and right hips, left and right pin bones, tail head, sacral, and cervical vertebrae. These locations remain constant as body condition score of a cow changes. The Euclidean distance between these keypoints was used to generate biometric features for each cow. The keypoint detection model was trained using 3,928 images over 900 epochs with a batch size of 14 and validated using 391 images. This trained model was used to predict keypoints on a different set of 41 cows over 5 d, resulting in a total of 6,944 images which correspond to the dataset for cow identification. The Random Forest (RF) algorithm for cow identification was trained with the first 4 d (5,376 images), and the last day served as the test set (1,568 images). For the cow identification, a RF algorithm was employed on the Euclidean distance of the keypoints. The hyperparameters were determined through a grid search on the training set, utilizing 5-fold cross-validation. The hyperparameters ‘mtry’ (which ranged from 1 to 10) and ‘ntree’ (which ranged from 250 to 500, in increments of 50) were optimized. For the second objective, due to the completely new environment, the keypoint model was fine-tuned using transfer learning for 240 epochs with a batch size of 14, utilizing 310 images. The model predicted keypoints in 1,593 images from different beef-on-dairy animals. This dataset was used to train the BW prediction model. The Euclidean distances from the predicted images and sex were used as features. A RF model was trained using a leave-one-animal-out cross-validation approach. The hyperparameters were determined through a grid search on the training set, using 10-fold cross-validation. The ‘mtry’ parameter ranged from 2 to 16, in increments of 2. For animal identification, the model achieved an accuracy, precision, recall, and F1-score of 92.7%, 89.0%, 90.2%, and 92.7%, respectively. To predict BW, the RF model achieved an R² of 0.86 and a root mean squared prediction error of 36.9 kg, representing 6.7% of the observed values average. These results suggest that keypoints located on the dorsal body surface can identify and weigh individual animals, even those lacking distinct coat color patterns.

Early wildfire detection using different machine learning algorithms

Early detection of wildfires is essential for mitigating their impact on forests and surrounding areas. In this study, we propose a wireless sensor node system that combines multiple low-cost sensors with an artificial intelligence-based detection method for early wildfire detection. The system architecture includes temperature, humidity, and smoke sensors, as well as a wireless communication module. Four machine learning classifiers, including decision trees, random forests, support vector machines, and k-nearest neighbors, were evaluated for their effectiveness in predicting wildfire detection using a dataset collected in a forest area. The results showed that the random forest algorithm with optimum hyperparameters had the highest accuracy in classifying fire and non-fire samples (77.95% correctly classified). The proposed system provides an effective and cost-efficient solution for early wildfire detection in large forest areas.

192 Prediction of fecal starch content of fattening cattle using near-infrared spectroscopy and machine learning

Abstract Starch content in cattle feces can be used to predict starch digestibility. Near-infrared spectroscopy (NIRS) can be used to rapidly and non-chemically predict fecal starch content after a calibration model has been consequentially developed. Recent advancements in machine learning have introduced potent methods for developing optimized models. In addition, diverse preprocessing methods, such as scatter correction and smoothing, are available to eliminate irrelevant variance sources. This study aimed to identify optimal spectral preprocessing and regression methods for predicting fecal starch content using NIR spectra. NIR spectra were measured for 196 fecal samples collected from 9 fattening cattle farms. Fecal starch content on a DM basis was measured using enzymatic reactions and colorimetry. The 196 samples were divided into a calibration set (n = 123), a validation set (n = 29) comprising four farms, and an external validation set (n = 44) comprising five farms that differed from the calibration set. Based on the calibration set, two preprocessing methods were selected using the partial least squares (PLS) model. In method 1, the calibration models were developed by a grid search 5-fold cross validation on the 1,152 preprocessed calibration sets, and preprocessing with the lowest root mean square error (RMSE) of cross-validation among these calibration sets was selected. In method 2, the original calibration set (n = 123) was divided into a calibration set (n = 116) and a preprocessing selection set (n = 7) with different neutral detergent fiber and similar starch contents. After preprocessing, calibration models were developed as described in method 1. The preprocessing selection sets were then predicted by the developed models, and preprocessing with the lowest RMSE among these preprocessing selection sets was selected. Using the preprocessing methods selected in methods 1 and 2, the validation and external validation sets were predicted using seven regression models. As a result, the preprocessing selected by method 1 was in the wavelength range of 400–2500 nm, with robust normal variate (RNV) scatter correction and first-order derivative Savitzky-Golay filtering (2nd order polynomial, 2-nm window size). The preprocessing selected by method 2 had a wavelength range of 1500–2500 nm, RNV scatter correction, and 2nd-order derivative Savitzky-Golay filtering (3rd order polynomial, 25-nm window size). Among the seven regression models, the Lasso model was the most accurate for prediction. The Lasso models based on the preprocessing methods selected by methods 1 and 2 had RMSE of external validation = 2.054% of DM and 1.540% of DM, respectively. The feature importance of the Lasso model based on preprocessing method selected by methods 2 was higher at wavelengths of 1775 nm and 2325 nm. In conclusion, seven models were developed and evaluated based on diverse preprocessing methods using NIRS. The Lasso model exhibited the highest accuracy.