Search Results (6,829)

Formal undergraduate research training programs are typically two years in length to support student development over time. However, some evidence suggests that many benefits of research participation can be achieved in a shorter duration, raising the question of whether shorter models can be as effective. This prospective study compared outcomes for two concurrent cohort-based research training models: a two-year Scholars Program and an accelerated one-year Fellows Program. The Fellows Program was designed for advanced students who joined the training program later than their peers for a variety of reasons. Both programs provided students with financial support, faculty-mentored research, a Learning Community, and other resources for professional development. The longer duration allowed Scholars to report significantly higher levels of research skills and produce more publications and professional presentations than Fellows by the end of their respective training programs. In contrast, the programs did not differ on graduation GPA, number of awards, or graduate school enrollment. Fellows also did not significantly differ from Scholars on psychosocial measures, including science/researcher identity, belongingness, cultural compatibility, family support, or time management. Findings suggest that a high-intensity, one-year model can deliver many broad academic and social benefits associated with the two-year model, while a longer two-year model provides students with more time and additional research opportunities to be more productive. We conclude that both types of programs should be available to best meet students’ needs and to make the most effective use of program resources. Full article

While drone delivery has gained significant scholarly and industrial interest, rural residents’ acceptance of these systems remains underexplored, despite the region’s acute logistics challenges and service inequities. Using survey data from rural U.S. residents, this study first estimates an ordered logistic regression (OLR) model to identify factors associated with five-level drone delivery acceptance. The study then compares OLR, multinomial logistic regression (MNL), Random Forest (RF), XGBoost, and LightGBM under matched feature sets to evaluate whether nonlinear machine-learning models improve prediction beyond the interpretable statistical baseline. Open-ended responses are coded into LLM-derived sentiment labels and added as supplementary predictors to test whether unstructured feedback improves acceptance prediction. Results show that willingness to pay is the strongest predictor of acceptance, while equitable same-day delivery demand and post-pandemic attitude adjustment are also positively associated with higher acceptance. Household disability status and urban accessibility are not significant after adjustment. In the five-level analysis, OLR provides a strong ordinal baseline, while XGBoost and other tree-based models improve selected class-level prediction metrics. In the binary high-acceptance analysis, machine-learning models show stronger predictive performance, especially when structured predictors are combined with sentiment features. This study contributes to rural drone-delivery literature by linking service equity, perceived value, and LLM-derived sentiment within a comparable statistical and machine-learning framework for rural service design. Full article

Rough set methodology has gained broad acceptance as a potent mathematical apparatus for feature selection within data mining and machine learning. Yet, classical rough sets hinge on equivalence relations to partition the universe, thereby demanding strict reflexivity, symmetry, and transitivity conditions that are arduous to satisfy in realistic settings. Although fuzzy rough sets have been explored to mitigate this rigidity, the entropy-based uncertainty measures employed in fuzzy approximation spaces remain acutely sensitive to data quality and noise corruption, potentially inducing severe bias in feature evaluation. Moreover, the literature currently lacks noise-tolerant uncertainty measures capable of accommodating a controlled fraction of classification errors while safeguarding the discriminative strength of feature subsets. Inspired by these gaps, this study develops a feature selection framework grounded in variable precision fuzzy entropy within the fuzzy rough set context. To this end, fuzzy decision is adopted to portray the membership degree of samples relative to decision classes, thereby enabling more precise detection and elimination of redundant attributes during approximation. An uncertainty quantifier termed fuzzy relational entropy is then introduced to appraise the distinguishing power of fuzzy similarity relations generated by attribute subsets. Leveraging fuzzy decision, a portfolio of uncertainty measure variants, specifically the variable precision joint discriminant index, the variable precision conditional discriminant index, and the variable precision mutual discriminant index, is developed to counteract noisy data effects. These variable precision discriminant indexes sanction a regulated error proportion and afford a measure of noise resistance. Finally, knowledge reduction for fuzzy decision systems is attacked from the angle of discriminative capability preservation, and a heuristic feature selection algorithm is crafted around the variable precision conditional discriminant index. Evaluation on twelve public UCI datasets reveals that the proposed algorithm effectively prunes redundant features and delivers competitive results against three representative alternatives: classical rough set, neighbourhood-based discriminant index, and fuzzy rough set feature selection. Additionally, it sustains stable classification performance across an extensive sweep of the variable precision parameter. Full article

Static, one-time authentication mechanisms such as passwords and PINs are increasingly inadequate for protecting mobile devices throughout an active session. Behavioral biometric continuous authentication (BBCA) addresses this gap by passively monitoring user-specific interaction patterns—keystroke dynamics, touch and swipe gestures, gait, and motion—to verify identity on an ongoing basis. This systematic review synthesizes 80 studies selected via a PRISMA-compliant protocol from IEEE Xplore, ACM Digital Library, Scopus, ScienceDirect, Web of Science, and SpringerLink (2017–2025). We examine behavioral and multimodal biometric modalities, machine learning approaches ranging from classical classifiers to deep sequence and transformer architectures, and their integration with intelligent personal agents, wearable devices, and IoT/edge infrastructures. Security analyses cover spoofing, adversarial and generative attacks, mimicry, and model-level threats including membership inference and reconstruction. Privacy-preserving mechanisms—cancelable biometrics, Bloom filter encodings, zero-knowledge proof protocols, federated learning, and blockchain-based identity management—are evaluated against practical trade-offs in energy consumption and latency on resource-constrained devices. Key research gaps are identified: the absence of standardized adversarial benchmarks, lack of end-to-end pipeline evaluations under simultaneous adversarial and privacy threat models, and limited user-centered studies on consent and acceptance of privacy-preserving mechanisms under frameworks such as GDPR. Recommended future directions combine adaptive multimodal fusion, privacy-preserving cryptography, energy-aware modality selection, and interdisciplinary human-centered evaluation to advance practical, resilient continuous authentication for mobile and assistant-enriched environments. Full article

With the rapid development of the Industrial Internet of Things, data-driven deep learning has achieved remarkable success in bearing fault diagnosis. However, traditional static models suffer from catastrophic forgetting when facing continuously emerging fault categories and limited edge storage. Existing class-incremental learning frameworks expose critical limitations when applied to 1D vibration signals on micro edge devices, including feature space oscillation, difficulty in anchoring lightweight classifiers, and prototype drift over long incremental cycles. To address these challenges, this paper proposes a novel end-to-end class-incremental fault diagnosis method based on lightweighting and feature calibration tailored for severe memory-constrained conditions. Specifically, a lightweight feature extraction mechanism based on an $L_2$ constraint is introduced to replace computationally expensive similarity distillation, effectively suppressing feature space oscillations and providing stable spatial coordinates for old knowledge. Moreover, a mandatory balanced center–margin hybrid replay (CAHM) strategy is designed to balance class representation while proportionally retaining class center prototypes and marginal hard examples, balancing the anchor accuracy of the Nearest Class Mean (NCM) classifier and the discriminability of the decision boundary. Furthermore, an ultra-low-cost linear prototype calibration module is constructed using a learnable affine transformation to actively redirect shifted old class prototypes with negligible inference latency. Extensive long-tail incremental experiments on the CWRU bearing dataset demonstrate that the proposed method forms a highly synergistic anti-forgetting closed loop. Under an extremely limited memory budget ($K=40$), the proposed framework achieves an outstanding final average accuracy of 98.92% after five incremental stages, significantly outperforming mainstream baselines such as iCaRL, PRIL, and SCKD and exhibiting exceptional robustness for continuous online monitoring on industrial edge devices. Full article

Objective: This study evaluated the educational value and motivational impact of unit-based Key Terminology Tests, combined with a dedicated glossary, in a second-year dental physiology course designed to support preparation for computer-based testing (CBT) and the Japanese National Dentist Examination. Methods: Seventy-six second-year dental students completed periodic terminology tests aligned with lecture units (50 items per test, 25 min each). Bonus points were awarded for scores of ≥45/50. At the end of the semester, students completed an anonymous questionnaire assessing perceived test burden (scope, duration, and number of items), preparation time, clarity and use of the glossary, perceived contribution to lecture comprehension, and the motivational impact of bonus points. Quantitative data were summarized descriptively, and free-text responses were analyzed thematically. Results: Most students rated the test scope and number of questions as appropriate (50.0% and 67.1%, respectively), although 48.7% perceived the scope as somewhat excessive or excessive, and 35.6% felt the time limit was short or somewhat short. The glossary improved lecture comprehension for 68.4% of respondents, and bonus points enhanced learning motivation for 53.9%. In addition, 86.9% reported using the glossary at least occasionally during routine study. Conclusions: Students perceived the terminology test system as supportive of lecture comprehension, perceived learning support, and sustained learning motivation, suggesting that it may serve as a useful preparatory approach for CBT and the National Dentist Examination. Full article

To address the problem that hotel recommendations in tourism activities do not consider the spatial relationship between hotels and scenic spots and the cost of tour routes, we construct a tourism hotel recommendation model based on ISTING-AGNES machine learning and an IDFST optimal route algorithm. Firstly, a scenic spot spatial clustering model based on the ISTING-AGNES machine learning algorithm is constructed, including a scenic spot ISG spatial topological model based on the neighborhood cell growth algorithm and an ISTING-AGNES machine learning algorithm based on the scenic spot ISG spatial topological model, which can realize spatial dimension reduction in tourist cities and construct tourism sub-regions for recommending scenic spots and hotels. Secondly, taking the tourism sub-regions as the modeling scope, a tourism hotel recommendation model based on the IDFST optimal route algorithm is constructed in which a closeness model between the tourists’ interests and the attributes of scenic spots in the sub-region is established to recommend the most matched scenic spots for tourists. Then, based on the recommended scenic spots, a tourism sub-interval optimal route algorithm based on IDFST and a tourism hotel recommendation model based on the optimal route decision forest algorithm are established to search for the global optimal tour route—with hotels as the starting and ending points and scenic spots as nodes—and to recommend the hotel with the most cost-effective tour route for tourists. The experiments prove that the constructed algorithm can output the hotel with the best geospatial location and the lowest tour route cost. Under the experimental conditions, compared with the hotel recommended by the weighted centroid positioning algorithm, the cost optimization rate reaches 5.98%. Compared with the greedy mountain climbing algorithm and the greedy BFS algorithm, the cost optimization rates reach 14.73% and 16.03%, proving that the constructed algorithm is feasible and advantageous over the traditional algorithms. Full article

The coating process of a ship-hull interior requires automation owing to significant occupational hazards associated with its working environment. The interior of a ship hull is large and structurally complex, and hence requires a highly redundant manipulator for such automation. This study proposed a control framework for redundancy resolution based on a pre-generated End-Effector(EE) reference path for an 8 Degree-of-Freedom (DoF) planar manipulator, used for coating automation. The proposed method employed tools such as Reinforcement Learning (RL) and Null Space Projection-based Reinforcement Learning (NSP-based RL). In RL, the action directly specifies the joint angular velocities, whereas in NSP-based RL, the NSP objective function’s gradient vector is generated. RL- and NSP-based RL share the same reward function and observation space. To mitigate the reward dominance and divergence issues that can arise in learning-based approaches, this study incorporated sequential sub-goal tracking and path planning information into the state definition. The proposed methods were evaluated in cluttered environments and compared with conventional NSP approaches. Experimental results showed that RL achieved superior obstacle avoidance, while NSP-based RL produced smoother and more consistent EE trajectories and velocities across all targets. Full article

Contemporary cyber threats increasingly employ multi-stage and behaviourally adaptive strategies that challenge static intrusion detection and non-adaptive deception mechanisms. Existing approaches typically treat threat detection, deception deployment, and adversarial attribution as separate functions, limiting timely response and underusing the behavioural evidence generated during attacker interaction. This study develops and evaluates a theory-informed computational and operational framework for autonomous cyber deception. The principal research artefact is a reusable closed-loop architecture rather than a single predictive model: it specifies the interacting components, interfaces, data and control flows, decision rules, and feedback mechanisms that connect detection, deception, telemetry, and attribution. Methodologically, the study follows an engineering design-and-evaluation approach comprising problem and requirement identification from the literature, architectural synthesis, component-level mathematical modelling, prototype implementation, and controlled cyber-range evaluation. In this context, modelling refers to the distinct computational models embedded within the framework: a hybrid detection model combining supervised classification, anomaly detection, and temporal sequence analysis; a Markov Decision Process and reinforcement-learning policy model for selecting and reconfiguring deception actions under engagement, intelligence-gain, resource, and containment objectives; and similarity-based and Bayesian attribution models for estimating MITRE ATT&CK techniques from incomplete behavioural evidence. The component models were developed offline using the NSL-KDD, CICIDS2017, UNSW-NB15, and ToN-IoT datasets, while the integrated prototype was evaluated separately in a controlled enterprise-like cyber range using reconnaissance, brute-force, exploitation, and multi-stage attack scenarios. The reported classification metrics were calculated from the labelled cyber-range evaluation events, not by pooling the four benchmark datasets. On this integrated cyber-range evaluation set, the system achieved 95.4% detection accuracy, 93.6% precision, 94.7% recall, and a 94.1% F1-score, with a mean detection latency of 85 ms. It also achieved 100% honeypot deployment reliability, 92% dynamic reconfiguration success, 88% fingerprinting resistance, and attacker engagement durations of up to 280 s. The attribution component demonstrated end-to-end generation of ATT&CK-aligned technique hypotheses from deception-derived telemetry; however, the present archived evaluation does not support per-technique or baseline-comparative performance claims. These findings show that specialised models and operational services can be coordinated within a unified adaptive defence process, while also identifying the additional class-level and ablation evidence required for rigorous attribution validation. Full article

This paper proposes a novel three-layered software architecture for processing and analyzing sequences, medical images, and patient-reported outcomes (PROs). The aim is to provide a comprehensive end-to-end solution for patient diagnosis by integrating all major types of bioinformatics data. This approach leverages advanced machine learning algorithms, including decision trees, support vector machines, neural networks, and quantum neural networks, to enhance the efficiency and effectiveness of precision medicine. A graphical user interface was constructed to validate the suggested approach and present the experimental results. Full article

The floods triggered by the 29 October 2024 cut-off low in Valencia (Spain) expose a persistent challenge in disaster risk reduction: extensive meteorological and territorial data do not automatically become timely, trusted or actionable public guidance. This conceptual synthesis uses the Valencia event as a diagnostic case and reconstructs selected evidence on rainfall, hydrological escalation and alert timing to develop a data-driven framework spanning observation, modelling, impact assessment, communication, decision-making and post-event learning. Here, “data-driven” denotes an end-to-end governance and translation process, not the development of a new forecasting model. The framework integrates four dimensions: data governance, user-centred visualization, uncertainty communication and school-based education. It also introduces territorial translation as the link between impact forecasts and place-specific infrastructures, routines, vulnerabilities and responsibilities. Its novelty lies in connecting the Early Warnings for All pillars and impact-based, people-centred warning approaches with an explicit educational and territorial learning loop. Its practical contribution is a responsibility matrix, a minimum governance package and an implementation roadmap with indicators for latency, reach, comprehension and protective action. The framework is intended for adaptation, rather than statistical generalization, across Mediterranean and other fast-onset flood contexts. Improved forecasts remain necessary but insufficient: loss reduction requires interoperable records, accessible impact-based messages and inclusive educational programmes that convert scientific information into situated collective competence. Full article

As data analytics increasingly rely on machine learning models for forecasting, classification, and prediction, effective visualisation becomes essential for transforming model outputs into practical insight. Yet the ways these outputs are visualised, and the evidence supporting those designs, remain fragmented across domains. This paper presents a systematic literature review of visualising machine learning model outputs in data analytics, focusing on how predicted outputs are communicated to end-users alongside performance and uncertainty information, and how these visual systems are evaluated in practice. Following PRISMA, we screened 330 articles from ACM Digital Library, IEEE Xplore, and PubMed and included 88 peer-reviewed studies published between Jan 2015 and July 2024. Across the corpus, we identify (1) recurring visual encoding and interaction patterns for interpreting predictions in temporal, spatio-temporal, and event-based settings; (2) common strategies for presenting model validation, calibration, and uncertainty; and (3) a wide range of evaluation approaches, from informal expert feedback to controlled user studies and deployments. The synthesis highlights persistent gaps in rigorous and comparable evaluation, challenges in supporting diverse user goals and expertise levels, and practical constraints that arise in operational contexts. We conclude by distilling practical implications for designing and assessing predictive visualisations, as well as outlining recommendations for future research and practice, with particular attention to improving uncertainty communication, strengthening evaluation rigour and comparability, and adopting evaluation methods that better reflect operational data analytics practice. Full article

This study analyses the evolution of the financial literature on climate risk, examining the integration of artificial intelligence techniques into its measurement and management. To this end, a bibliometric approach is employed based on 221 articles indexed in the Web of Science Core Collection, using the Bibliometrix package. Moving beyond existing descriptive bibliometric reviews on ESG and green finance, the novelty of this paper lies in its analytical focus on how financial science operationalises quantitative AI mechanisms to price and integrate climate transition risk into asset and portfolio valuation. The structural analysis reveals that natural language processing (NLP) and digital transformation acting as driving motor themes, suggesting that the reviewed literature associates AI innovation policies with the mitigation of corporate greenwashing and enhance information transparency. Furthermore, while machine learning algorithms establish the cross-cutting predictive foundation for risk assessment, empirical evidence unveils a critical academic shift of traditional ‘financial performance’ towards a declining quadrant, indicating that empirical studies frequently find that that multi-phase investments in risk technologies do not yield immediate financial returns. Finally, the study maps a persistent geographical gap where emerging markets lack the data infrastructure of advanced economies, alongside isolated high-dimensional causal econometric niches like double machine learning. This analytical mapping provides key implications for global risk management and future quantitative research avenues. Full article

Rapid and accurate multi-class fault diagnosis is essential for high-end equipment because different fault categories require different maintenance responses. This study aims to develop a lightweight and discriminative diagnostic framework that can identify multiple fault categories from non-stationary vibration signals while reducing redundant computation. High-frequency vibration signals provide direct condition information, but long sequences, noise, nonlinear dynamics, and non-stationary behavior make raw-signal classification unreliable. From an entropy-based information-processing perspective, the key issue is to separate informative fault modes from redundant fluctuations and enlarge inter-class distinctions in the probabilistic decision space. This study proposes Fastformer, an integrated framework for vibration-based fault identification. Empirical Mode Decomposition first converts each signal into Intrinsic Mode Functions to reduce modal mixing and preserve fault-related oscillatory components. The resulting components are processed by an encoder-oriented Q/K/V dot-product scoring mechanism, which constructs compact spatiotemporal embeddings without adopting a complete Transformer architecture. Validation-guided pruning removes low-contribution attention responses, while a Margin-Enhanced Fault Softmax classifier optimized with a cross-entropy-based objective strengthens category separation. By combining stable decomposition, lightweight attention scoring, pruning, and probabilistic margin learning, Fastformer achieves faster and more stable convergence. On the XJTU-SpurGear dataset, Fastformer obtains precision, recall, F1-score, and AUC values of 1.000. Additional validation on the HUST bearing dataset further shows that Fastformer achieves the best overall performance among the compared methods, with an AUC value of 0.9596. Full article

We present a neuroadaptive control framework applied to a tactile sensing interface for real-time, human-guided physical interaction with a robotic arm. The framework employs a dual-loop architecture consisting of an inner-loop neuroadaptive controller that compensates for nonlinear robot dynamics and an outer-loop ARMA–RLS tactile mapping that converts tactile sensor voltages into planar end-effector displacement commands. Four piezoresistive tactile sensors mounted on the robot end-effector are calibrated individually using autoregressive moving-average (ARMA) models updated through recursive least squares (RLS). The proposed tactile interface does not estimate an absolute Cartesian force/torque wrench; instead, it learns a user- and sensor-specific voltage-to-motion command mapping for planar guidance. To evaluate robustness to user variability, 28 participants completed the calibration experiments, producing 112 user- and sensor-specific calibration models. The calibration procedure achieved millimeter-level displacement-prediction accuracy, with a mean RMSE of approximately 2.70 mm across participants. After calibration, participants used the tactile interface to guide the robot along a predefined figure-eight trajectory. The average nearest-path tracking error decreased from $11.07\pm 5.45$ mm in the initial trial to $8.41\pm 3.48$ mm in the final trial, indicating improved tactile-guided path following after repeated exposure to the interface. During these experiments, the inner neuroadaptive controller maintained bounded joint-space tracking errors. Overall, the proposed calibration and control framework provides a low-cost physical interface for planar human-guided robot motion without requiring a wrist-mounted force/torque sensor. Full article