833 papers
This study leverages a stacking ensemble of LightGBM and XGBoost models with an ExtraTrees meta-learner on the 2025 Global Findex dataset to predict digital payment adoption with 93.04% accuracy, revealing that education, internet access, and income are the most critical drivers of financial inclusion.
This systematic literature review synthesizes fragmented research on AI in digital banking to propose the novel AIDBRM 2.0 framework, which integrates customer experience, fraud detection, and risk management through cross-cutting mechanisms like a Unified Continuous Trust Score, a Generative AI Governance Sandbox, and Federated Cross-Institution Learning to address existing gaps in explainability and architectural alignment.
This paper presents a systematic literature review of 173 studies on retrieval enhancement techniques for Retrieval-Augmented Generation, categorizing them by technical contribution and pipeline stage to identify key gaps in evaluation and efficiency while proposing a structured research agenda for future advancements.
This paper proposes AJAD-BoostNet, a novel hybrid framework combining ResNet50-based deep feature extraction, PCA dimensionality reduction, K-Means clustering, and XGBoost boosting to achieve efficient and robust histopathological classification of lung and cervical cancers, demonstrating superior performance over traditional CNNs in resource-constrained settings.
This study introduces the framework of "Machine Psychology" to reveal that aligned large language models, while free of explicit racial bias, exhibit ten novel, highly structured identity-reactive behaviors in response to ethnically-cued user names, necessitating a shift from basic harm mitigation to comprehensive behavioral evaluation.
This study systematically evaluates five open-source lexico-grammatical tagging systems for multidimensional analysis, revealing that while they successfully replicate general register distinctions, their resulting dimension scores vary significantly due to inconsistent feature operationalization rather than differences in underlying NLP pipelines.
This paper proposes a four-layer framework that enhances vision-only counter-UAV systems by converting tracked trajectories into structured motion evidence and combining them with key-frame mosaics in a vision-language model to generate interpretable behavior labels, urgency estimates, and natural-language rationales, demonstrating that structured trajectory data is the primary driver of accurate UAV behavior analysis.
This paper proposes a two-stage path fitting framework that integrates a "minimum toll" principle with a multi-dimensional anomaly detection engine to address gantry data issues in expressway tolling, achieving a 2.1% accuracy improvement and a 70% reduction in billing deviations while introducing a novel dual-indicator evaluation model to ensure result reliability.
The paper introduces CHONN, a modular, circuit-inspired framework that employs Kirchhoff-based cascade composition to create stable, interpretable high-order neural dynamics, thereby outperforming traditional depth-stacking approaches in solving partial differential equations and enhancing visual perception tasks.
The paper introduces 4DHemoX, a physics-informed AI framework leveraging a hybrid dataset and a Navier-Stokes-embedded neural architecture to bridge the sim-to-real gap and enable reliable, generalizable, and data-efficient cerebrovascular hemodynamic profiling for clinical applications.