Search Results (459)

Facial emotion recognition (FER) in real-world settings is limited by the semantic mismatch between discrete emotion categories and continuous Valence–Arousal–Dominance (V-A-D) dimensions and the lack of reliable uncertainty estimates under incomplete facial evidence. Existing uncertainty-aware FER studies mainly address annotation ambiguity or training-time reliability, leaving the behavior of predictive uncertainty under progressive input degradation insufficiently examined. This paper proposes BGDC (Bayesian Gaussian-mixture Distributional Consistency), a multi-task FER framework that integrates a GMM-based soft consistency module with a context-conditioned Bayesian regression head and explicitly models aleatoric and epistemic uncertainty. To evaluate predictive reliability, a controlled masking protocol is introduced to remove facial information under different spatial configurations. On FER2013-VAD, BGDC attains the highest classification accuracy of 0.6943 and the highest mean V-A-D CCC of 0.6079 among the compared configurations, and it yields a stronger epistemic uncertainty-error correspondence than MC Dropout in a single-model setting. Controlled masking further shows that the epistemic uncertainty of BGDC tracks task-relevant facial information loss rather than masking ratio alone: it rises with regression error when diagnostically important regions are removed, and it contracts when the masked region is largely task-irrelevant. Combining Bayesian uncertainty with the GMM-based distributional prior thus enables reliability-aware multi-task FER, in which controlled masking serves as a diagnostic intervention rather than as a benchmark of accuracy degradation alone. Full article

Background/Objectives: Inheritance classification in rare diseases remains challenging because curated knowledge is incomplete, heterogeneous, and imbalanced across inheritance categories. Disease-level inheritance modeling can support knowledge organization, annotation review, and hypothesis generation in rare-disease resources. This paper introduces RareFusion-Net, a multimodal benchmark framework for disease-level inheritance classification, and evaluates whether integrating ontology-enriched disease text, structured epidemiological metadata, and gene-association information improves prediction in curated rare-disease knowledge bases. RareFusion-Net is intended for knowledge modeling, not individual patient diagnosis. Methods: We developed RareFusionBalanced, a gated multimodal fusion model that combines biomedical disease descriptions, structured metadata, and gene-related information using auxiliary supervision. Ontology-enriched disease text was treated as the dominant semantic modality, while tabular and gene modalities were incorporated as complementary evidence when available. Robustness was improved using balanced regularization, selective transformer fine-tuning, dropout, weight decay, label smoothing, early stopping, and prediction aggregation across random seeds. Evaluation included accuracy, macro-F1, micro-F1, macro-AUC, mean average precision, calibration metrics, class-wise analysis, statistical testing, and ablation experiments. Results: RareFusionBalanced achieved 0.7382 test accuracy, 0.6284 macro-F1, 0.7382 micro-F1, 0.9183 macro-AUC, and 0.6686 mean average precision. Calibration was favorable, with an expected calibration error of 0.0395 and a Brier-OVR of 0.0528. The multimodal model slightly outperformed TextOnly-TransformerBalanced, but improvement over the best TF-IDF baseline was not statistically significant. Ablation showed ontology-enriched text as the strongest modality, with gene associations adding complementary value. Conclusions: RareFusion-Net provides a practical benchmark for ontology-aware rare-disease inheritance modeling. Results suggest selective multimodal benefit while highlighting minority-class difficulty, limited statistical superiority, need for external validation, and improved biological interpretability. Full article

The high dropout rate in massive open online courses (MOOCs) continues to limit their potential in promoting inclusive and sustainable learning. Although many prediction models have been used to identify potential dropouts, most studies view dropout as a static classification problem and fail to clearly reveal the dynamic trajectory of learner participation over time. Therefore, this study introduces a phased analysis perspective, treating MOOC dropout as a process that continuously evolves at different stages. On the basis of the KDDCUP2015 dataset, we constructed behavioral characteristics at three time points: the first week, the third week, and the fifth week. By combining robust feature analysis and interpretable models, we systematically examined the changing patterns of dropout modes. The results revealed significant differences across the different stages. In the early stage of the course, dropout was related mainly to the unstable interaction behaviors of learners, such as restricted access to resources and irregular participation rhythms. In the middle and late stages, task-oriented behaviors, especially those related to video-based learning activities, gradually became key factors. Notably, high-frequency video participation does not always reduce the risk of dropout; when video activity is high but the overall interaction rate is low, it is more likely to indicate an increase in the risk of dropout. These results indicate that the combination of behaviors is more crucial than mere activity levels. By revealing the changing characteristics of behaviors at different stages, this study helps support the design of more practical early warning methods. Full article

Background: The convergence of machine learning (ML), gamification, and critical thinking assessment within adaptive educational platforms has accelerated since 2020, driven by large language models (LLMs) and graph neural networks (GNNs). No prior systematic review has jointly addressed all three dimensions, and Central Asian educational contexts remain underrepresented. Methods: Following PRISMA 2020 guidelines, we searched Scopus (n  =  4396) and OpenAlex (n  =  4152) for publications from 2016 to 2026. Quality assessment used the Mixed Methods Appraisal Tool (MMAT 2018; threshold ≥  2), yielding 82 papers. Five research questions addressed ML personalization (RQ1), gamification and engagement (RQ2), critical thinking assessment tools (RQ3), recommendation algorithms (RQ4), and regional applicability in Kazakhstan and Central Asia (RQ5). Results: Transformer-based and GNN models dominate the recent literature (52% of corpus from 2025), with an accuracy of 91–97% for dropout prediction and learning path recommendation under single-institution conditions. Gamification studies report up to 90% student satisfaction; LLM-based critical thinking assessment shows promise but faces validity concerns. Thirteen papers address Central Asian contexts. Conclusions: Significant gaps persist: no integrated gamification–critical thinking framework exists, recommendation systems lack explainability, and Kazakh-language datasets are severely underrepresented. Future research should prioritize multilingual adaptive systems, explainable algorithms, and privacy-preserving federated learning for low-resource contexts. Full article

Student dropout remains an important challenge in higher education because it affects degree completion, institutional resource efficiency, workforce preparation, and students’ long-term socioeconomic opportunities. This requires not only accurate predictions but also decision support that is both reliable and aware of uncertainty. This study posits that the amalgamation of probabilistic modeling, uncertainty quantification, and graph-based refinement can augment both predictive reliability and decision support for the early detection of dropouts. A reliability-centered predictive framework is presented, integrating Educational Competition Optimization (ECO)-based feature selection, probabilistic Support Vector Classification (SVC), isotonic regression for probability calibration, and split conformal prediction for distribution-free uncertainty quantification. In addition, a similarity-driven Graph-based Fuzzy Cellular Automata (Graph-FCA) refinement mechanism is developed, where student relationships are modeled using a k-nearest neighbor graph with radial basis function similarity. Entropy-based confidence weighting is used to control uncertainty-aware propagation. An Explainable Artificial Intelligence layer based on SHAP provides both global and local interpretability, and fairness-aware evaluation assesses consistency across demographic groups. The suggested framework maintains predictive performance while improving probabilistic reliability. The Graph-FCA refinement achieves an accuracy of 0.7503, which is close to the calibrated ECO–SVC baseline (Accuracy = 0.7537; Macro-F1 = 0.6704) and also reduces the Brier score. The conformal prediction layer achieves empirical coverage close to the desired confidence level, ensuring reliable uncertainty estimates. The ECO–SVC–Conformal–GraphFCA framework transforms traditional classification into a reliable, understandable, and uncertainty-aware early warning system, enhancing its usefulness for ethical and informed decision-making in engineering education. Full article

The in-plane shear performance of carbon fiber-reinforced polymer (CFRP) composites is critical for structural design but is challenged by significant property scatter. This study aims to achieve individualized prediction of the complete shear stress–strain curve for each composite specimen using only a single early-stage digital image correlation (DIC) strain field. Systematic in-plane shear tests were conducted on 45 laminated carbon fiber/epoxy specimens with synchronized full-field DIC data and macroscopic load–displacement records. A lightweight encoder–decoder convolutional neural network was developed, taking a single DIC strain contour map at 0.2% global strain as input and mapping it directly to the full-range stress–strain curve up to failure for that specific specimen. Data augmentation and Dropout regularization mitigated the small-sample challenge. The proposed model achieved strong predictive performance across the five-fold cross-validation yielded a mean R² of 0.926 ± 0.022 and a mean RMSE of 6.37 ± 1.14 MPa for stress. Individual specimen predictions on the test set yielded an average R² of 0.945, with a minimum of 0.821, confirming robust capability across scattered properties. Residual analysis elucidated error characteristics across deformation stages. This research provides a novel paradigm for non-destructive, early-stage individualized assessment of composite mechanical properties, with applications in structural health monitoring and probabilistic design. Full article

Accurate photovoltaic (PV) power forecasting is pivotal for facilitating the integration of renewable energy into modern power systems and supporting sustainable energy development. However, existing methods often rely on single deep learning architectures, require complex preprocessing, suffer from training instability, and lack the ability to capture long-range temporal dependencies. To address these issues, this study develops and compares two hybrid deep learning models—ConvTempNet and DilaTransNet—for hourly PV energy prediction using meteorological and temporal data from two Portuguese PV stations. Quantitative results show that the optimized ConvTempNet achieves superior hourly predictive accuracy with an hourly RMSE of 1.16 kWh and an R² of 0.95 at Tartaruga (2.66 kWh, R² = 0.95 at Zarco). Systematic evaluations were conducted, including dropout ablation (a systematic test of different dropout rates to assess model robustness and regularization effects) (0.2–0.4), performance assessment using RMSE, R², MAE, and MAPE, and sensitivity analysis to assess predictive accuracy and variable importance. Results show that the optimized ConvTempNet yields superior hourly accuracy with an hourly RMSE = 1.16 kWh and an R² = 0.95 at Tartaruga (2.66 kWh, R² = 0.95 at Zarco). The tuned DilaTransNet shows stronger robustness to moderate dropout. Solar radiation is the dominant input variable, while temperature, humidity, and hour affect the two models differently. The two models exhibit complementary strengths, supporting site-specific parameter optimization for reliable PV forecasting. Full article

Objectives: Alzheimer’s disease (AD) remains one of the most prevalent neurodegenerative conditions among older adults, underscoring the urgent need for accurate and ethically grounded early detection methods. Artificial intelligence (AI) techniques, particularly machine learning and deep learning models, show promise in leveraging neuroimaging biomarkers to support early diagnosis. However, significant challenges persist regarding model explainability, accountability, and responsible implementation in real-world healthcare settings. This study presents a generalized Responsible AI (RAI) framework composed of four core components—explainability, fairness, predictive performance, and uncertainty quantification—to address these challenges. Method: Using the TADPOLE neuroimaging dataset, we implemented a Feedforward Neural Network (FNN) within a unified Responsible AI (RAI) framework integrating explainability, fairness, predictive performance, and uncertainty quantification. Although Random Forest achieved slightly higher predictive accuracy (95%), the FNN was selected as the primary model because it better supports end-to-end uncertainty estimation through Monte Carlo Dropout, enabling more reliable clinical decision support. Results: The proposed framework demonstrated strong predictive performance (92% accuracy), improved fairness reflected by an equalized odds difference of 0.124, and progressively lower predictive entropy across training iterations, indicating enhanced confidence in predictions. The framework further enabled model transparency through explainability analyses and supported the identification of low-confidence predictions for potential clinical review. Conclusions: Our findings highlight not only the feasibility of integrating RAI principles into AD prediction pipelines but also the persistent challenges of applying such frameworks to real-world clinical data. This work contributes practical insights toward operationalizing Responsible AI in healthcare contexts. Full article

Elevated concentrations of fine particulate matter ($P{M}_{2.5}$) are a critical threat to respiratory health worldwide. Therefore, there is an urgent need for precise urban forecasting systems for public health management. Technological advancements in the domains of continuous environmental monitoring and deep learning have enabled large-scale data acquisition, processing, and modeling. Existing predictive models typically depend on auxiliary meteorological inputs, which are frequently inaccessible within standard ground-level monitoring networks. Furthermore, conventional approaches often fail to adequately capture the complex spatio-temporal interactions of pollutants. To address these limitations, this study presents the Spatio-Temporal Endogenous Attention Network (STEA-Net), a forecasting framework designed to operate exclusively without weather variables. Validated on a comprehensive multi-year historical dataset (Jan 2015–Feb 2020) from diverse monitoring stations in India, STEA-Net employs a hybrid adjacency matrix that integrates physical geographical distances with functional clustering to accurately map pollutant transport pathways. Utilizing this structural map, a Graph Attention Network dynamically evaluates the spatial influence of neighboring nodes, while a Bidirectional LSTM processes the underlying temporal sequences. Experimental results demonstrate that STEA-Net substantially surpasses traditional machine learning algorithms and provides competitive performance against advanced deep learning baselines. The proposed model achieves a peak Coefficient of Determination ($R^2$) of 0.9294 (5-seed average: $0.9273\pm 0.0023$) and a peak RMSE of 14.38 µg/m³ (5-seed average: $14.59\pm 0.23$ µg/m³), effectively adapting to the dynamic volatility of urban pollution levels. The model exhibits architectural stability with a Monte Carlo dropout verified deviation of $\pm 2.22$ µg/m³. This research provides a forecasting architecture that retains competitive predictive performance under the strict operational constraint of meteorology-free deployment in resource-constrained urban monitoring environments. Full article

Medical text classification is an important task in biomedical natural language processing, but prediction errors remain problematic in high-stakes settings where reliability matters in addition to accuracy. To address this challenge, this paper proposes a learned selective deferral framework for biomedical sentence classification that allows uncertain predictions to be deferred under constrained review budgets. The framework combines a transformer-based classifier with uncertainty estimation, temperature scaling, and a learned deferral policy that predicts the likelihood of model error from multiple signals, including confidence, entropy, calibration-aware features, and Monte Carlo Dropout descriptors. Deferral decisions are applied under fixed budgets to improve the use of limited review capacity. Experiments on the PubMed 200k RCT dataset show that budget-constrained deferral reduces system-level risk. Using PubMedBERT as the primary backbone, deferring 20% of the highest-risk cases reduces system risk from 0.1108 to 0.0360. Compared with a calibrated confidence-threshold baseline, the learned policy provides modest but generally favorable improvements, with statistical significance observed at the 20% budget. Additional experiments across PubMedBERT, BioBERT, and SciBERT suggest that the framework transfers across biomedical transformer backbones, while calibration improves the reliability of confidence estimates and learned policies outperform random deferral. Full article

This paper investigates the tracking control problem for a class of networked nonlinear systems in a non-ideal communication environment, where both internal communication constraints (delays and packet dropouts) and external random deception attacks are taken into account. From a symmetry perspective, the backward and forward channels constitute a paired sensing–actuation structure, and channel-dependent imperfections may destroy their functional coordination. To compensate for the resulting sensing–actuation mismatch, a data-driven sliding-mode predictive tracking control scheme is developed without relying on an explicit system model. First, an equivalent dynamic linearization is adopted to represent the input–output behavior using a data-dependent incremental model. Then, using delayed measurements together with historical input–output data, an online estimator is constructed to update the pseudo partial derivative (PPD). Based on the estimated PPD, a multi-step predictor is further designed to generate the predicted outputs, and a data-driven sliding-mode predictive tracking controller is proposed by imposing a discrete reaching law on the predicted outputs. Rigorous analysis is provided to ensure the stability of the closed-loop system and to guarantee that the tracking error remains bounded, together with an explicit bound that reveals the influence of the delay horizon, estimation mismatch, and attack amplitudes. Finally, numerical simulations under square-wave and sinusoidal references validate the effectiveness and robustness of the proposed approach. Full article

Multivariate time-series anomaly detection on physical sensor networks faces three challenges that generic deep learning models inadequately addressed: heterogeneous sensor reliability, context-dependent anomaly scoring, and inactionable binary outputs lacking per sensor attribution. We propose an uncertainty-aware Temporal Convolutional Network (TCN) framework built on two tightly integrated uncertainty-driven components: (i) an Adaptive Uncertainty-Aware Attention (AUAA) mechanism that gates temporal attention weights by per sensor predictive uncertainty obtained from Monte Carlo dropout; and (ii) a Dynamic Weight Adapter that learns context-sensitive blending of reconstruction error and uncertainty via a GRU over weight history. The architecture also includes an exploratory per sensor attribution head, which we audit rather than claim: a controlled-perturbation test shows it is not yet causally faithful. We complement the empirical architecture with two distribution-free theoretical results: a Chebyshev-type false-positive bound on the hybrid anomaly score, and a Monte Carlo posterior moment convergence result at rate $O\left(M^{-1/2}\right)$. Evaluated on four-month indoor air quality sensor data, the Full Enhanced model achieves $R^2=0.9988$ and MSE $1.65\times {10}^{-4}$, a $25.2\%$ MSE reduction over the Base TCN ($R^2=0.9984$, MSE $2.20\times {10}^{-4}$). Because the IAQ stream is unlabeled, the primary quantitative detection evaluation uses the labeled Skoltech Anomaly Benchmark (SKAB), a publicly available industrial water-circulation corpus disjoint from the IAQ training distribution; it yields an 8.8 × F1 advantage ($0.477$ vs. $0.054$) and a 14.4 × recall advantage ($0.418$ vs. $0.029$) for the proposed model configuration over the Base TCN at a validation-calibrated threshold applied without retuning. Against twelve established detectors under a unified protocol, the proposed model attains the best F1 and recall, while the strongest reconstruction baselines retain higher precision and a marginally higher ROC-AUC, a recall-driven trade-off. Ablation isolates each component’s contribution, the detector degrades gracefully under channel masking and noise, and the distribution-free false-positive bound is empirically respected. The framework retains a low inference cost ($0.16$ ms per window at $M=20$ Monte Carlo samples, including the uncertainty pass). Full article

Data leakage is a major source of overoptimistic performance estimates in machine learning-based predictive modelling. In higher education, dropout models are increasingly used to support interventions, yet leakage may arise not only from technical mistakes but also from an underspecified prediction task. This article conceptualizes predictive modelling as a temporally specified decision configuration and proposes Time-Consistent Dropout Prediction (TCDP) as a diagnostic framework for longitudinal educational data. TCDP defines four core components: an explicit prediction cutoff, a pre-cutoff information set, a risk-set-consistent target population, and a temporally appropriate validation design. The validation component comprises two design conditions: entity-level train–test separation and cohort-level temporal consistency. A structured methodological screening of recent dropout-prediction studies shows that temporal anchoring is becoming more common, whereas risk-set definition and validation hierarchy remain less consistently formalized. The empirical demonstration uses a pseudonymized longitudinal student panel from a Hungarian higher education institution, accompanied by a controlled synthetic reproduction package. Results show that row-wise validation in data with repeated student observations generates increasing train–test entity overlap across prediction cutoffs and inflates AUC, especially for high-capacity or instance-based models. Enforcing entity-level separation removes this inflation and yields lower, more realistic performance ranges. The paper contributes an operational validation grammar linking prediction time, admissible information, population eligibility, and validation design. Full article

(1) Background: Student burnout, widely regarded as a form of “hidden dropout” among adolescents, is associated with lower educational quality and mental health. Grounded in the Study Demands–Resources (SD–R) and Conservation of Resources (COR) theories, this study investigates the relationship between school-based resources, family dynamics, and personal resources by examining how teacher emotional support is associated with burnout through family cohesion and meaning in life; (2) Methods: a moderated mediation model was tested using a sample of 1224 adolescents (Mage = 14.27, SD = 1.72; 48% female); (3) Results: Analysis revealed that: 1. Teacher emotional support significantly and negatively predicted student burnout (β = −0.28, p < 0.001). 2. Family cohesion partially mediated this relationship, accounting for 36% of the total effect. 3. Meaning in life significantly moderated both the direct path and the second half of the mediation pathway (family cohesion → burnout). Notably, meaning in life was associated with a stronger negative association between teacher emotional support and student burnout, but a weaker negative association between family cohesion and student burnout, a pattern consistent with differential resource utilization; (4) Conclusions: These findings suggest a differentiated pattern of resource interplay: school-based emotional resources may connect to family-based relational resources, and the protective role of each external resource may be further moderated by adolescents’ internal meaning systems. These findings highlight the agentic role of adolescents in resource management and point to the value of multi-system interventions. Full article

Drought is a slowly evolving, multi-driver hydro-meteorological hazard whose accurate early prediction is a cornerstone of climate-smart agriculture and water-resource planning. Existing data-driven drought forecasting frameworks suffer from three persistent limitations: (i) most models concatenate heterogeneous climate variables into a single flat feature vector, implicitly assuming a single dominant driver such as precipitation, even though atmospheric moisture demand, radiation and wind-mediated evapotranspiration co-determine drought onset; (ii) wavelet preprocessing is typically applied to the full series, introducing future-information leakage that violates the operational causality requirement of forecasting; and (iii) most architectures predict a single horizon and provide no causal attribution explaining when, where and which climatic variables initiated the event. This study proposes WaveDroughtNet, a multi-modal, multi-horizon deep-learning framework that addresses these limitations through five integrated components: (a) a strictly causal Daubechies-4 wavelet decomposition computed in a rolling fashion; (b) six modality-specific encoders with stochastic modality dropout (p = 0.15); (c) cross-modal multi-head attention with four heads; (d) a four-layer temporal convolutional network (TCN) backbone with dilation factors yielding a 240-step receptive field; and (e) a post hoc DroughtOriginTracer that combines temporal attention, modal-attribution and inter-district propagation scans. The Standardised Precipitation Evapotranspiration Index (SPEI), used as the supervisory target, is computed following the canonical Vicente-Serrano formulation. water balance $D=P-\mathrm{P}\mathrm{E}\mathrm{T}$ (Hargreaves PET) at a 4-week (≈1-month) timescale, fitted with a three-parameter log-logistic distribution via L-moments, validated by Kolmogorov–Smirnov goodness-of-fit testing ($\alpha =0.05$) per district, and standardised through the inverse-normal cumulative distribution function. Trained on 18,304 weekly district records from NASA POWER reanalysis (2014–2025) covering all 32 districts of Tamil Nadu, India, WaveDroughtNet uses only 256,869 parameters and produces, in a single forward pass, four forecasts (1 week, 1 month, 3 months, 1 year). On the held-out 2024 test partition ($N=1728$), the model attains weighted $F1=0.9221$ and $R^2=0.8512$ at the 1-week horizon, and weighted $F1=0.8498$ and $R^2=0.6812$ at the 1-year horizon. Diebold–Mariano tests confirm that WaveDroughtNet significantly outperforms naive persistence, seasonal naive, LSTM, ConvLSTM and a vanilla Transformer at the 3-month and 1-year horizons (p < 0.001). The DroughtOriginTracer successfully back-projects 15 Coimbatore events to causal origins 29–41 weeks prior to onset. We explicitly acknowledge three limitations that constrain operational deployment in its current form—zero severe events in the 2024 test partition ($F{1}_{severe}$ = 0.000), static inter-district modelling, and absence of vegetation-index supervision—and propose concrete mitigation pathways in the Discussion. Full article