Learning Model Research Articles

Spatial Prediction of Soil Continuous and Categorical Properties Using Deep Learning Approaches for Tamil Nadu, India

Large-scale mapping of soil resources can be crucial and indispensable for several of the managerial applications and policy implications. With machine learning models being the most utilized modeling technique for digital soil mapping (DSM), the implementation of model-based deep learning methods for spatial soil predictions is still under scrutiny. In this study, soil continuous (pH and OC) and categorical variables (order and suborder) were predicted using deep learning–multi layer perceptron (DL-MLP) and one-dimensional convolutional neural networks (1D-CNN) for the entire state of Tamil Nadu, India. For training the deep learning models, 27,098 profile observations (0–30 cm) were extracted from the generated soil database, considering soil series as the distinctive stratum. A total of 43 SCORPAN-based environmental covariates were considered, of which 37 covariates were retained after the recursive feature elimination (RFE) process. The validation and test results obtained for each of the soil attributes for both the algorithms were most comparable with the DL-MLP algorithm depicting the attributes’ most intricate spatial organization details, compared to the 1D-CNN model. Irrespective of the algorithms and datasets, the R2 and RMSE values of the pH attribute ranged from 0.15 to 0.30 and 0.97 to 1.15, respectively. Similarly, the R2 and RMSE of the OC attribute ranged from 0.20 to 0.39 and 0.38 to 0.42, respectively. Further, the overall accuracy (OA) of the order and suborder classification ranged from 39% to 67% and 35% to 64%, respectively. The explicit quantification of the covariate importance derived from the permutation feature importance implied that both the models tried to incorporate the covariate importance with respect to the genesis of the soil attribute under study. Such approaches of the deep learning models integrating soil–environmental relationships under limited parameterization and computing costs can serve as a baseline study, emphasizing opportunities in increasing the transferability and generalizability of the model while accounting for the associated environmental dependencies.

Determination of Organic Carbon Content of the Soils within the Greenhouses Built on Pyroclastic Deposits in Isparta Settlement Area

Soil organic carbon (SOC) is an important indication of soil health and helps to sustain soil fertility. As a result, determining its composition and the factors that influence it is critical for long-term soil nutrient management, especially in controlled conditions such as greenhouses. This study utilizes machine learning to classify SOC content in greenhouses built on pyroclastic deposits in the Isparta region. A dataset of 276 samples and eight variables—clay (%), silt (%), sand (%), soil electrical conductivity (EC), pH, elevation, slope, and aspect—were used to model SOC values. SOC content was classified into five classifications: very low (2.3%). In this study, five machine learning models—Logistic Regression (LR), K-Nearest Neighbors (KNN), Support Vector Machine (SVM), Decision Tree (DT), and Random Forest (RF)—were evaluated using cross-validation to determine their classification accuracy, precision, recall, F-score, and ROC area. Random Forest (RF) and Decision Tree (DT) outperformed the other models, with RF achieving the highest overall accuracy (76.4%), precision (77.3%), and AUC (0.904), followed by DT at 75.4% and AUC of 0.874. This study shows the practicality of machine learning models in categorizing SOC content, highlighting their importance for long-term soil health and fertility control in greenhouse conditions. To improve model efficacy, future studies should include more auxiliary variables, such as soil physical and chemical qualities and lithological data, as well as a wider range of soil types.

Detection and Classification of Novel Attacks and Anomaly in IoT Network using Rule based Deep Learning Model

Detection and Classification of Novel Attacks and Anomaly in IoT Network using Rule based Deep Learning Model

Differentiated Learning Models: Students’ Learning Style, Readiness, and Interest in Learning Regarding the Results of Fiqh

<em>The differentiation learning model is a novel approach that caters to the demands of the 21st- century education system. Studies conducted by Indonesian scholars on this model are scarce, with limited focus on its impact on students' cognitive abilities and metacognitive awareness. Therefore, this study seeks to investigate the impact of incorporating a differentiated learning approach founded on the learning preferences, preparedness, and engagement of pupils on Fiqh's academic performance. An explanatory survey technique employing the Stratified Random Sampling methodology is utilized in this research. The sample comprises all fifth-grade students of the State Islamic Primary School in Tulungagung Regency, amounting to a total of 507 learners. The study's sample comprises four randomly selected State Islamic Primary Schools from the seven Tulungagung Regency, with a total of 276 students studying Fiqh. The data collection instruments used include test sheets for learning outcomes, notes on creative thinking skills, and motivation questionnaires. The data is then analyzed using a two-way Analysis of Variance (ANOVA) with a significance level of 5%. The results of the research indicate that: (1) There is no discernible disparity in Fiqh learning outcomes based on learning style. (2) There is no disparity in Fiqh learning outcomes based on students’ learning readiness. (3) Disparities in Fiqh learning outcomes based on students’ learning interests exist. (4) A noteworthy interaction exists between the implementation of differentiation learning aspects of learning styles, learning readiness, and students’ learning interests on Fiqh learning outcomes. The findings of this study suggest that the differentiation learning model provides an alternative approach to learning that addresses the needs of the modern era.</em>

Estimating ambient population density using physical features from GIS and machine learning: a study based on Japanese neighborhood

ABSTRACT This research aimed to develop an alternative method for estimating ambient population density at the neighborhood scale by utilizing a simple and universally available dataset. Physical environment data from GIS & OSM databases, basic statistics, and population density data from Mobile Spatial Statistics were combined to train tree-based Machine Learning models. The experiment resulted in an XGBoost model using 16 features capable of estimating ambient population density across three classes of outcome with 75.9% accuracy. The trained models were analyzed and visualized using SHAP and Partial Dependence Plot techniques to reveal feature importance and their threshold values. The study concludes that physical features can be effectively used as predictors of ambient population density and highlights areas for further investigation.

Abstract B024: Deciphering the neurogenic tumor microenvironment: A novel gene network driving innervation in metastatic melanoma offers prognostic and therapeutic insights

Abstract Melanoma, a highly aggressive and life-threatening cancer, poses a significant therapeutic challenge, particularly for patients with distant metastasis, who face a 5-year survival rate of only approximately 22.5%. This underscores the urgent need for innovative therapeutic approaches. Emerging evidence suggests that sensory and parasympathetic innervation within the tumor microenvironment (TME) plays a crucial role in driving metastasis and immune evasion, partly through the promotion of M2 macrophages and cancer stem cells. To uncover the mechanisms behind tumor-specific innervation in metastatic melanoma, we employed a systems biology approach integrating bulk RNAseq from diverse melanoma cell lines, TCGA-SKCM data analysis, single-cell RNAseq, and spatial transcriptomics. This comprehensive analysis led us to identify a previously uncharacterized gene regulatory network (GRN) orchestrating tumor innervation. This GRN is driven by axon guidance cues (Netrin, Slit, Semaphorin), their receptors (DCC, Robo, Neuropilin, ITGB1), and transcription factors (RAS, RND1, TFAP2, CAMKII). We further discovered that the pioneer factor FOXC2, known for its association with metastasis, plays a critical role in opening the promoters of innervation-specific genes. Additionally, we identified Ephrin and CXCR4-mediated crosstalk between neurons and tumor cells. Machine learning models validated the accuracy of this GRN in predicting melanoma metastasis, and a neural network-based model confirmed the enrichment of nerve cells near metastatic tumor cells within the TME in TCGA immunohistochemistry data. The functional relevance of this network was further supported by CRISPR-mediated knockout studies, which disrupted tumor-specific signaling pathways. Notably, we also observed the enrichment of genes linked to mRNA processing bodies and neuron-specific membrane-less condensates in metastatic melanoma. These findings highlight the critical role of neuronal infiltration in melanoma metastasis. Currently, therapeutic strategies targeting this aspect of the TME remain limited. Our identified gene set offers potential as a metastatic predictor, and targeting the neural component of the TME represents a promising new therapeutic avenue for melanoma. Citation Format: Anjana Goli, Subhajit Dutta. Deciphering the neurogenic tumor microenvironment: A novel gene network driving innervation in metastatic melanoma offers prognostic and therapeutic insights [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Tumor-body Interactions: The Roles of Micro- and Macroenvironment in Cancer; 2024 Nov 17-20; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2024;84(22_Suppl):Abstract nr B024.

Predicting brain tumor presence using machine learning models

Predicting brain tumor presence using machine learning models

PRITrans: A Transformer-Based Approach for the Prediction of the Effects of Missense Mutation on Protein–RNA Interactions

Protein–RNA interactions are essential to many cellular functions, and missense mutations in RNA-binding proteins can disrupt these interactions, often leading to disease. To address this, we developed PRITrans, a specialized computational method aimed at predicting the effects of missense mutations on protein–RNA interactions, which is vital for understanding disease mechanisms and advancing molecular biology research. PRITrans is a novel deep learning model designed to predict the effects of missense mutations on protein–RNA interactions, which employs a Transformer architecture enhanced with multiscale convolution modules for comprehensive feature extraction. Its primary innovation lies in integrating protein language model embeddings with a deep feature fusion strategy, effectively handling high-dimensional feature representations. By utilizing multi-layer self-attention mechanisms, PRITrans captures nuanced, high-level sequence information, while multiscale convolutions extract features across various depths, thereby enhancing predictive accuracy. Consequently, this architecture enables significant improvements in ΔΔG prediction compared to traditional approaches. We validated PRITrans using three different cross-validation strategies on two newly reconstructed mutation datasets, S315 and S630 (containing 315 forward and 315 reverse mutations). The results consistently demonstrated PRITrans’s strong performance on both datasets. PRITrans demonstrated strong predictive capability, achieving a Pearson correlation coefficient of 0.741 and a root mean square error (RMSE) of 1.168 kcal/mol on the S630 dataset. Moreover, its robust performance extended to independent test sets, achieving a Pearson correlation of 0.699 and an RMSE of 1.592 kcal/mol. These results underscore PRITrans’s potential as a powerful tool for protein-RNA interaction studies. Moreover, when tested against existing prediction methods on an independent dataset, PRITrans showed improved predictive accuracy and robustness.

Improving the Learning Outcomes of Class VIII Students Through the Implementation of the Project-Based Learning (PjBL) Model in Science Learning

This research aims to improve the learning outcomes of junior high school students in science learning.This research is a type of classroom action research carried out in one of the public junior high schools in the Madiun district. The study population was class VIII with class VIII E samples. The data analysis technique used was to compare learning outcomes in the pre-cycle to the next cycle. The results of the analysis showed that the project-based learning (PjBL) learning model was able to improve the cognitive learning outcomes of junior high school students. The success indicator of the study was seen in the improvement in cognitive learning outcomes of students who achieved a minimum completeness criteria (MCC) score of 75. While the tang performance indicator used in this study is the success rate of cognitive learning outcomes up to 75%, in the pre-cycle, as many as 71.43% of the total students have not reached the specified minimum completeness criteria (MCC). While the remaining 38.10% have reached minimum completeness criteria (MCC). The implementation of the project-based learning (PjBL) learning model shows an increase in student success in learning. In cycle I, the percentage of student success reached 61.9%. In the second cycle, the percentage of student success reached 76.1%, so there was an increase of 33.4%.

Pengaruh Model Pembelajaran Inkuiri Terbimbing Terhadap Hasil Belajar Siswa Pada Mata Pelajaran Informatika Di SMK

The learning process has an influence in achieving significant learning outcomes, including learning media, learning motivation and learning models, learning models are a learning plan that will take place which contains a learning design. The learning model refers to the learning method that will be used, including learning objectives, learning stages, learning environment, and classroom management. To improve learning outcomes, it is necessary to improve the model that influences learning outcomes, where the learning model greatly influences learning outcomes, this can improve learning outcomes significantly. In the learning that took place at the Antarctic 1 Sidoarjo school, there were students who were hyperactive and less conducive due to several things including students who did not like to be restrained and needed more attention to be able to implement conducive learning, in overcoming this problem, the researcher chose a guided inquiry learning model to improve learning outcomes, this is because this learning model is more fun with learning that is packaged in a question, which is able to improve students' understanding by asking questions to satisfy their curiosity, in implementing this model the researcher used the One Group Pretest Posttest research design, by giving the same treatment to a population with two meetings, starting with giving recognition to the Pretest and ending with the Posttest. Then the data is processed to see the effect of giving treatment in the population. Hypothesis testing using the paired t-test formula (T-Paired) produces a decision that is rejected H0, the calculation of the influence with the eta square test from the table of criteria results shows a large influence. It can be concluded that there is an influence on student learning outcomes in informatics subjects using the guided inquiry learning model.

A Study on Enhancing Crop Yield Prediction with Accuracy through Machine and Deep Learning Models

Accurate crop yield prediction is essential for optimizing agricultural resource allocation and supporting farmers in making informed crop management decisions. This study investigates the effectiveness of various machine learning and deep learning models for predicting crop yields based on key agricultural parameters, including Nitrogen (N), Phosphorous (P), Potassium (K), temperature, humidity, pH, and rainfall. We apply multiple algorithms—Logistic Regression, Multilayer Perceptron, Sequential Minimal Optimization (SMO), J48, Random Forest, REP Tree, and a proposed deep learning model—to evaluate their predictive accuracy in classifying agricultural items. Each model’s performance is assessed using metrics such as Kappa statistic, Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), Relative Absolute Error (RAE), Root Relative Squared Error (RRSE), and the time taken to build the model. Findings reveal that while traditional machine learning models like Random Forest and REP Tree show robust performance, the proposed deep learning model demonstrates enhanced accuracy and efficiency, making it highly promising for crop yield prediction applications. This study underscores the potential of advanced learning algorithms to improve agricultural productivity by providing actionable insights for crop planning and resource management.

Predicting the compressive strength of high-performance concrete using an interpretable machine learning model.

Reliable predictions of concrete strength can reduce construction time and labor costs, providing strong support for building construction quality inspection. To enhance the accuracy of concrete strength prediction, this paper introduces an interpretable framework for machine learning (ML) models to predict the compressive strength of high-performance concrete (HPC). By leveraging information from a concrete dataset, an additional six features were added as inputs in the training process of the random forest (RF), AdaBoost, XGBoost and LightGBM models, and the optimal hyperparameters of the models were determined using 5-fold cross-validation and random search methods. Four interpretable ML models for predicting the compressive strength of HPC, including the RF, AdaBoost, XGBoost and LightGBM models, which combine feature derivation and random search, were constructed. In addition, the SHapley Additive exPlanations (SHAP) method was applied to analyze the effects of the input features on the prediction results of the LightGBM model, which combines feature derivation and random search. The results showed that input features such as age, water/cement ratio, slag, and water were the key influences for predicting the compressive strength of HPC. Input features such as the superplastic/cement ratio, slag/cement ratio, and ash/cement ratio had nonsignificant impacts on the predicted compressive strength.

Machine learning prediction of brain metastasis invasion pattern on brain MRI scans

Abstract Background Brain metastasis invasion pattern (BMIP) is an emerging biomarker associated with recurrence-free and overall survival in patients, and differential response to therapy in preclinical models. Currently, BMIP can only be determined from the histopathological examination of surgical specimens, precluding its use as a biomarker prior to therapy initiation. The aim of this study was to investigate the potential of machine learning (ML) approaches to develop a non-invasive magnetic resonance imaging (MRI)-based biomarker for BMIP determination. Methods From an initial cohort of 329 patients, a subset of 132 patients met inclusion criteria for this retrospective study. We evaluated the ability of an expert neuroradiologist to reliably predict BMIP. Thereafter, the dataset was randomly divided into training/validation (80% of cases) and test subsets (20% of cases). Ground truth for BMIP was histopathologic evaluation of resected specimens. Following MRI sequence co-registration, advanced feature extraction techniques deriving hand-crafted radiomic features with traditional ML classifiers and convolution-based deep learning (CDL) models were trained and evaluated. Different ML approaches were used individually or using ensembling techniques to determine the model with the best performance for BMIP prediction. Results Expert evaluation of brain MRI scans could not reliably predict BMIP, with an accuracy of 44-59% depending on the semantic feature used. Among the different ML and CDL models evaluated, the best performing model achieved an accuracy of 85% and an F1-score of 90%. Conclusion ML approaches can effectively predict BMIP, representing a non-invasive MRI-based approach to guide management of patients with brain metastases.

Developing a jam-absorption strategy for mixed traffic flow at signalized intersections using deep reinforcement learning

ABSTRACT Jam-absorption driving (JAD) can effectively prevent the generation and propagation of traffic oscillation. To alleviate the traffic congestion in the signalized intersection with mixed traffic flow, including human driving vehicles (HDVs) and connected and automated vehicles (CAVs), this study provides a jam-absorption driving strategy based on the traffic delay prediction of the mixed platoon under traffic congestion. An online traffic congestion prediction method with the objective of JAD is proposed and focuses on the leaving state of the trajectory to achieve fast capture of congestion features. Then, with real-time status and prediction information, we develop a Jam-absorption driving strategy based on a deep reinforcement learning (DRL) model to improve adaptability to the mixed traffic environment. The results show that this strategy can suppress more than 70% of traffic oscillations with excellent execution efficiency, improving traffic safety and efficiency.

Machine learning model based on SERPING1, C1QB, and C1QC: A novel diagnostic approach for latent tuberculosis infection

AbstractBackgroundLatent tuberculosis infection (LTBI) is a significant source of active tuberculosis (ATB), yet distinguishing between them is challenging because specific biomarkers are lacking.MethodsWe analyzed four microarray datasets (GSE19491, GSE37250, GSE54992, GSE28623) from the gene expression omnibus to identify differentially expressed genes (DEGs). Using protein–protein interaction (PPI) networks and LASSO‐SVM algorithms, we selected three candidate biomarkers and evaluated their diagnostic efficacy. The expression levels of core genes were validated by RNA sequencing of healthy, ATB, and LTBI groups in a real‐world cohort. We conducted Gene Ontology and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses, predicted shared upstream miRNAs, constructed miRNA–hub and transcription factor (TF)–hub gene networks, and performed immune infiltration analysis.ResultsThree hub genes (SERPING1, C1QC, C1QB) were identified from 45 DEGs by PPI networks and machine learning screening. The diagnostic model based on the three hub genes had an area under the curve (AUC) value of 0.843 in the training set GSE19491 and 0.865 in the validation set GSE28623. Real‐world transcriptome sequencing confirmed the expression trends of the hub genes across healthy, LTBI, and ATB groups. GO analysis showed that the 45 hub genes were primarily associated with immune inflammatory responses and pattern recognition receptors, whereas KEGG analysis indicated enrichment in complement and coagulation cascades. The miRNA–hub and TF–hub gene network analysis identified nine miRNAs and the zinc finger TF GATA2 as potential co‐regulators of SERPING1, C1QC, and C1QB. Immune cell infiltration analysis identified significant differences in the immune microenvironment between LTBI and ATB, with macrophages and natural killer cells showing significant correlations with tuberculosis infection.ConclusionThe diagnostic model with SERPING1, C1QC, and C1QB shows promise in distinguishing LTBI from ATB, indicating its potential as a diagnostic tool.

Enhancement of Breast Cancer Classification Using Bat Feature Selection with Recurrent Deep Learning

DNA is a valuable tool for classifying expression of genes in detection of breast cancer. Gene expression data are biological data that extract valuable hidden information from gene datasets. Extracting useful features from datasets is a challenging task. Our gene expression dataset had a small number of samples but many features. This paper compared three types of recurrent deep learning models, including recurrent neural networks (RNN), long short-term memory (LSTM), and gated recurrent unit (GRU), for classification of breast cancer. The goals of the study were to improve the accuracy of classification and to enhance the effectiveness of feature selection; the basic principle was to select the best features from the original datasets. The bat algorithm assists in selecting the best relevant feature when integrated with recurrent deep learning models, which improves breast cancer classification by leveraging training datasets. Data preprocessing involves removing unnecessary columns and filling out missing values with the median value. The result was a comparative study using recurrent deep learning with the bat algorithm to classify breast cancer. The bat algorithm with LSTM achieved higher accuracy than RNN and GRU, where GRU had the lowest accuracy.

Variance‐driven security optimisation in industrial IoT sensors

AbstractThe Industrial Internet of Things (IIoT) has transformed industrial operations with real‐time monitoring and control, enhancing efficiency and productivity. However, this connectivity brings significant security challenges. This study addresses these challenges by identifying abnormal sensor data patterns using machine learning‐based anomaly detection models. The proposed framework employs advanced algorithms to strengthen industrial defences against cyber threats and disruptions. Focusing on temperature anomalies, a critical yet often overlooked aspect of industrial security, this research fills a gap in the literature by evaluating machine learning models for this purpose. A novel variance‐based model for temperature anomaly detection is introduced, demonstrating high efficacy with accuracy scores of 0.92 and 0.82 on the NAB and AnoML‐IOT datasets, respectively. Additionally, the model achieved F1 scores of 0.96 and 0.89 on these datasets, underscoring its effectiveness in enhancing IIoT security and optimising cybersecurity for industrial processes. This research not only identifies security vulnerabilities but also presents concrete solutions to improve the security posture of IIoT systems.

EXPRESS: From Social to Purchase: Customer Selection in Social Group Buying

Social scope group buying has emerged as an increasingly popular promotional strategy and has served as a new customer acquisition tool. In the service industry, companies use social group buying (SGB) to recruit new customers and promote full-price products. Through SGB activities, customers can trade their social capital to form buying groups, experience SGB-offered sample products, and further alleviate uncertainty regarding expensive full-price products before making a final purchase. We investigate this novel SGB phenomenon by examining customers’ decisions throughout the “experience-conversion” process. In collaboration with a leading online educational platform, we examine customers’ grouping behavior during SGB activities and analyze their subsequent purchases. Our analysis reveals an interesting pattern in which non-grouped customers have a higher proportion of full-price product purchases than grouped customers. We postulate that, in addition to observations from operational data, unobserved social information is important for gaining a deeper understanding of the customer behaviors underlying this pattern. Employing a Bayesian learning framework, we model customers’ three-stage discrete-choice decision-making processes and quantify two influential social information factors: social cost and social learning. By incorporating social information, our Bayesian learning model demonstrates improved effectiveness in assisting companies with accurately predicting final purchases in the conversion process. We provide actionable insights into how companies can employ our model to strategically tailor SGB designs by customer segments to increase overall purchase rates. Our study sheds light on the significance of social information in enhancing customer management and refining SGB design.

Predicting cell type-specific epigenomic profiles accounting for distal genetic effects.

Understanding how genetic variants affect the epigenome is key to interpreting GWAS, yet profiling these effects across the non-coding genome remains challenging due to experimental scalability. This necessitates accurate computational models. Existing machine learning approaches, while progressively improving, are confined to the cell types they were trained on, limiting their applicability. Here, we introduce Enformer Celltyping, a deep learning model which incorporates distal effects of DNA interactions, up to 100,000 base-pairs away, to predict epigenetic signals in previously unseen cell types. Using DNA and chromatin accessibility data for epigenetic imputation, Enformer Celltyping outperforms current best-in-class approaches and generalises across cell types and biological regions. Moreover, we propose a framework for evaluating models on genetic variant effect prediction using regulatory quantitative trait loci mapping studies, highlighting current limitations in genomic deep learning models. Despite this, Enformer Celltyping can also be used to study cell type-specific genetic enrichment of complex traits.

PENGEMBANGAN MODEL PEMBELAJARAN TEMATIK INTEGRATIF BERBASIS KEARIFAN LOKAL UNTUK MEMBENTUK KARAKTER ANAK (PENELITIAN PENGEMBANGAN DI RA KHALIFAH INSANI METRO

This research was motivated by problems discovered during initial observations in Group A RA. Khalifah Insani Metro City, namely some students do not understand their respective characters because their learning activities are still conventional so they rarely link local culture with learning themes. The aim of this research is to develop a prototype of an integrative thematic learning model based on local wisdom for early childhood and determine the effectiveness of an integrative thematic learning model based on local wisdom in forming character. The research approach used in this research is Research and Development (R&D) with J. Moonen's developmental instructional model. The sample used was 1 teacher and 20 group A students at RA Khalifah Insani Metro City. The results of the research show that the development of an integrative thematic learning model based on local wisdom to shape children's character regarding product use is able to facilitate and instill good manners in students and can create appropriate and effective learning.