Search Results (186)

Personalized e-learning seeks to adapt sequences of learning activities to individual learners, yet most existing adaptive platforms continue to rely on heuristic rules or single-objective optimization strategies. This paper introduces AlphaLearn, a conceptual evolutionary agent that frames learning pathway design as a constrained multi-objective optimization problem. The framework integrates knowledge graphs, learner modelling, and evolutionary algorithms to generate, evaluate, and iteratively refine candidate learning pathways under multiple pedagogical criteria. The contribution of this work is threefold. First, it presents a structured architectural framework for evolutionary learning pathway optimization, including a formal description of the optimization cycle and pathway representation. Second, it provides a descriptive analysis of large-scale learning analytics data from the Open University Learning Analytics Dataset (OULAD), illustrating substantial variability in learner outcomes, failure rates, and dropout across modules. Third, it offers an explicit discussion of fairness and bias mitigation, positioning equity as an integral dimension of adaptive pathway optimization rather than a post-hoc concern. The descriptive findings highlight pronounced heterogeneity in learner performance and engagement, motivating the need for adaptive systems capable of balancing learning effectiveness, efficiency, engagement, and fairness. While AlphaLearn is presented as a conceptual and methodological framework rather than a validated system, it establishes a foundation for future empirical evaluation and the development of fairness-aware evolutionary approaches to personalized e-learning. Full article

Malicious traffic detection in the Industrial Internet of Things (IIoT) faces significant challenges, primarily due to the scarcity of labeled data, high inference latency on resource-constrained edge devices, and the lack of comprehensibility in deep learning models. To overcome these limitations, this paper proposes MeeDet, a novel detection framework that integrates Mamba-based state-space modeling, a dynamic early-exit mechanism, and Large Language Model (LLM)-driven comprehensibility. The proposed MeeDet operates through a four-stage pipeline. First, raw packet captures are preprocessed into header-only, standardized stride-based sequences. Second, a 12-layer unidirectional Mamba backbone is pretrained on unlabeled data using two complementary tasks: Masked Byte Modeling for byte-level semantics and Next-Flow Prediction for long-range flow-level temporal coherence. Third, the model is fine-tuned by attaching lightweight binary heads to each Mamba layer, allowing for the early termination of high-confidence benign samples and adaptive routing of ambiguous flows to deeper layers. Finally, for detected malicious samples, structured prompts containing key network traffic features are processed by an LLM to generate human-readable diagnostic reports, without affecting real-time detection latency. Extensive experiments on five public IIoT datasets demonstrate the superiority of MeeDet over existing baselines. MeeDet achieves F1-scores exceeding 0.98 on key benchmarks while significantly reducing computational overhead. Specifically, at a 1% malicious traffic ratio, MeeDet requires only 1.7 MFLOPs and 1.58 ms of average inference latency, representing a reduction of over 70% in computational cost compared to strong pretrained baselines. Full article

Vision–language–action (VLA) models often suffer from limited robustness in long-horizon manipulation tasks—where robots must execute extended sequences of actions over multiple time steps to achieve complex goals—due to their inability to explicitly exploit structural symmetries and to react adaptively when such symmetries are violated by environmental uncertainty. To address this limitation, this paper proposes PI-VLA, a symmetry-aware predictive and interactive VLA framework for robust robotic manipulation. PI-VLA is built upon three key symmetry-driven principles. First, a Cognitive–Motor Synergy (CMS) module jointly generates discrete and continuous action chunks together with predictive world-model features in a single forward pass, enforcing cross-modal action consistency as an implicit symmetry constraint across heterogeneous action representations. Second, a unified training objective integrates imitation learning, reinforcement learning, and state prediction, encouraging invariance to task-relevant transformations while enabling adaptive symmetry breaking when long-horizon deviations emerge. Third, an Active Uncertainty-Resolving Decider (AURD) explicitly monitors action consensus discrepancies and state prediction errors as symmetry-breaking signals, dynamically adjusting the execution horizon through closed-loop replanning. Extensive experiments on long-horizon benchmarks demonstrate that PI-VLA achieves state-of-the-art performance, attaining a 73.2% average success rate on the LIBERO benchmark (with particularly strong gains on the Long-Horizon suite) and an 88.3% success rate in real-world manipulation tasks under visual distractions and unseen conditions. Ablation studies confirm that symmetry-aware action consensus and uncertainty-triggered replanning are critical to robust execution. These results establish PI-VLA as a principled framework that leverages symmetry preservation and controlled symmetry breaking to enable reliable and interactive robotic manipulation. Full article

Background: Colorectal cancer (CRC) is the third most diagnosed type of cancer and the second leading cause of cancer-related death. However, the increase in CRC incidence observed over the last 50 years has been accompanied by an overall reduction in mortality thanks to improved diagnostic strategies, patient follow-up, and more targeted therapies. Gastrointestinal adenomatous polyposis syndromes are a group of hereditary syndromes that predispose individuals to gastrointestinal tumors. These syndromes, characterized by the onset of gastrointestinal adenomas, are genetically heterogeneous. Methods: We analyzed 60 subjects with clinical suspicion or diagnosis of polyposis using next-generation sequencing (NGS). An additional 20 healthy individuals, all negative for pathogenic variants, were included in the study as a control population. We also performed bioinformatic analyses to investigate the hypothesis that benign variants could still be partially destructive, even though they cannot, by themselves, be responsible for the onset of disease. Results: Germline pathogenic variants were identified in 55% (33/60) of affected patients (MUT⁺), while variants of uncertain significance (VUS) were identified in 18.3% of affected patients (11/60). No variants were detected in the remaining 26.7% (16/60) of patients (MUT⁻). A genotype-phenotype correlation emerged from this study: MUT⁺ patients exhibited a significantly earlier age of onset and a higher number of polyps compared to VUS or MUT⁻ patients. Furthermore, Mendelian inheritance was significatively more frequent in MUT⁺ and VUS patients than in MUT⁻ individuals. Finally, the investigation of benign variants identified an SNP (single nucleotide polymorphism) of the APC gene promoter and a cluster of variants in POLD1, in which bioinformatic analysis predicted altered gene expression. Conclusions: These results suggest that, although MUT⁻ patients may develop multiple gastrointestinal adenomatous polyps, they are likely to have a familial predisposition rather than a Mendelian disorder. Furthermore, we propose that certain benign variants may be partially deleterious, potentially contributing to disease onset and/or act as phenotypic modifiers, likely through additive effects. Full article

Chronic myeloid leukemia (CML) has undergone a significant shift over the past two decades, transitioning from a fatal malignancy to a chronic, highly manageable disease with near-normal life expectancy for most patients. This transformation has been driven by the development of BCR-ABL1-targeted tyrosine kinase inhibitors (TKIs), which have enabled durable disease control and deep molecular responses (DMRs) in the majority of patients with chronic-phase CML. As long-term survival outcomes have plateaued across available agents, contemporary management has shifted beyond disease suppression toward optimizing long-term safety, quality of life, and the achievement of treatment-free remission (TFR). This review summarizes current evidence on molecular monitoring strategies, the comparative efficacy and toxicity profiles of first-, second-, and third-generation TKIs, and emerging advances in response assessment. Patient-centered TKI selection is discussed in the context of cardiovascular risk, comorbidities, treatment tolerability, and survivorship goals, reflecting the growing emphasis on individualized therapy in chronic-phase CML. Molecular monitoring strategies are examined in parallel, highlighting the clinical importance of early and sustained DMRs in guiding therapeutic decisions and TFR eligibility. Although RT-qPCR remains the standard for molecular monitoring, emerging high-sensitivity techniques such as digital droplet PCR and next-generation sequencing provide complementary value by improving the detection of low-level residual disease, refining risk stratification, and enabling earlier identification of resistance. Emerging therapeutic strategies and advances in response assessment further highlight ongoing efforts to enhance the depth and durability of remission while minimizing long-term toxicity. These developments support a more precise, individualized, and outcome-driven approach to modern CML management. Full article

Against the dual backdrop of the rural revitalization strategy and the pursuit of high-quality, balanced urban–rural education, optimizing rural campus spaces has emerged as an important lever for addressing educational resource disparities and improving pedagogical quality. However, conventional evaluation of campus space optimization faces two systemic dilemmas. First, top-down decision-making often neglects the authentic needs of diverse stakeholders and place-based knowledge, resulting in spatial interventions that lose regional distinctiveness. Second, routine public participation is constrained by geographical barriers, time costs, and sample-size limitations, which can amplify professional cognitive bias and impede comprehensive feedback formation. The compounded effect of these challenges contributes to a disconnect between spatial optimization outcomes and perceived needs, thereby constraining the distinctive development of rural educational spaces. To address these constraints, this study proposes a novel method that integrates regional spatial feature recognition with digital media-based public perception assessment. At the data collection and ethical governance level, the study strictly adheres to platform compliance and academic ethics. A total of 12,800 preliminary comments were scraped from major social media platforms (e.g., Douyin, Dianping, and Xiaohongshu) and processed through a three-stage screening workflow—keyword screening–rule-based filtering–manual verification—to yield 8616 valid records covering diverse public groups across China. All user-identifying information was fully anonymized to ensure lawful use and privacy protection. At the analytical modeling level, we develop a Transformer-based deep learning system that leverages multi-head attention mechanisms to capture implicit spatial-sentiment features and metaphorical expressions embedded in review texts. Evaluation on an independent test set indicates a classification accuracy of 89.2%, aligning with balanced and stable scoring performance. Robustness is further strengthened by introducing an equal-weight alternative strategy and conducting stability checks to indicate the consistency of model outputs across weighting assumptions. At the scenario interpretation level, we combine grounded-theory coding with semantic network analysis to establish a three-tier spatial analysis framework—macro (landscape pattern/hydro-topological patterns), meso (architectural interface), and micro (teaching scenes/pedagogical scenarios)—and incorporate an interpretive stakeholder typology (tourists, residents, parents, and professional groups) to systematically identify and quantify key features shaping public spatial perception. Findings show that, at the macro level, naturally integrated scenarios—such as “campus–farmland integration” and “mountain–water embeddedness”—exhibit high affective association, aligning with the “mountain-water-field-village” spatial sequence logic and suggesting broad public endorsement of ecological campus concepts, whereas vernacular settlement-pattern scenarios receive relatively low attention due to cognitive discontinuities. At the meso level, innovative corridor strategies (e.g., framed vistas and expanded corridor spaces) strengthen the building–nature interaction and suggest latent value in stimulating exploratory spatial experience. At the micro level, place-based practice-oriented teaching scenes (e.g., intangible cultural heritage handcraft and creative workshops) achieve higher scores, aligning with the compatibility of vernacular education’s “differential esthetics,” while urban convergence-oriented interdisciplinary curriculum scenes suggest an interpretive gap relative to public expectations. These results indicate an embedded relationship between public perception and regional spatial features, which is further shaped by a multi-actor governance process—characterized by “Government + Influencers + Field Study”—that mediates how rural educational spaces are produced, communicated, and interpreted in digital environments. The study’s innovative value lies in integrating sociological theories (e.g., embeddedness) with deep learning techniques to fill the regional and multi-actor perspective gap in rural campus POE and to promote a methodological shift from “experience-based induction” toward a “data-theory” dual-drive model. The findings provide inferential evidence for rural campus renewal and optimization; the methodological pipeline is transferable to small-scale rural primary schools with media exposure and salient regional ecological characteristics, and it offers a new pathway for incorporating digital media-driven public perception feedback into planning and design practice. The research methodology of this study consists of four sequential stages, which are implemented in a systematic and progressive manner: First, data collection was conducted: Python and the Octopus Collector were used to crawl online comment data related to Fuwen Township Central Primary School, strictly complying with the user agreements of the Douyin, Dianping, and Xiaohongshu platforms. Second, semantic preprocessing was performed: The evaluation content was segmented to generate word frequency statistics and semantic networks; qualitative analysis was conducted using Origin software, and quantitative translation was realized via Sankey diagrams. Third, spatial scene coding was carried out: Combined with a spatial characteristic identification system, a macro–meso–micro three-tier classification system for spatial scene characteristics was constructed to encode and quantitatively express the textual content. Finally, sentiment quantification and correlation analysis was implemented: A deep learning model based on the Transformer framework was employed to perform sentiment quantification scoring for each comment; Sankey diagrams were used to quantitatively correlate spatial scenes with sentiment tendencies, thereby exploring the public’s perceptual associations with the architectural spatial environment of rural campuses. Full article

This study aimed to determine the sexual maturation pattern of Shenxian pigs by combining observation, teaser boar testing, and back-pressure methods, and to apply this pattern for early breeding to shorten the generation interval and increase production efficiency. Subsequently, high-throughput transcriptome technology was used to compare gene expression levels in testicular tissues of Shenxian pigs before and after sexual maturity, as well as between sexually mature Shenxian pigs and Shenxian × Large White crossbred pigs. Functional analysis of differentially expressed genes (DEGs) was conducted to screen candidate genes related to sexual maturation and precocity in Shenxian pigs. The results showed that boars reached sexual maturity at an average age of 116 days in winter and 129 days in summer. For sows, the first estrus occurred at 114 days, the second at 134 days, and the third at 154 days in winter; corresponding ages in summer were 125, 144, and 164 days, respectively. The duration of estrus was around 3 days, and the estrus interval was approximately 20 days for both seasons. Comparative trials revealed no significant change in production performance when selection and first mating were conducted at 5 months of age compared to previous practices. Transcriptome sequencing of testicular tissues before and after sexual maturity in Shenxian pigs identified 6016 upregulated genes, primarily associated with reproduction and sperm function, influencing sexual maturation. The comparison between sexually mature Shenxian pigs and crossbred pigs identified 582 upregulated genes, mainly involved in hormone synthesis, affecting the onset of puberty in Shenxian pigs. After intersecting and functionally analyzing the upregulated genes from both sets, SRD5A1 and CYP11B2 were selected as the most likely candidate genes to affect precocious puberty in Shenxian pigs. Full article

With the continuous improvement of power system intelligence, multimodal data generated during distribution network maintenance have grown exponentially. However, existing power multimodal datasets commonly suffer from issues such as low sample quality, frequent factual errors, and inconsistent instruction expressions caused by regional differences.Traditional sample correction methods mainly rely on manual screening or single-feature matching, which suffer from low efficiency and limited adaptability. This paper proposes a multimodal sample correction framework based on large-model instruction enhancement and knowledge guidance, focusing on two critical modalities: temporal data and text documentation. Multimodal sample correction refers to the task of identifying and rectifying errors, inconsistencies, or quality issues in datasets containing multiple data types (temporal sequences and text), with the objective of producing corrected samples that maintain factual accuracy, temporal consistency, and domain-specific compliance. Our proposed framework employs a three-stage processing approach: first, temporal Bidirectional Encoder Representations from Transformers (BERT) models and text BERT models are used to extract and fuse device temporal features and text features, respectively; second, a knowledge-injected assessment mechanism integrated with power knowledge graphs and DeepSeek’s long-chain-of-thought (CoT) capabilities is designed to achieve precise assessment of sample credibility; third, beam search algorithms are employed to generate high-quality corrected text, significantly improving the quality and reliability of multimodal samples in power professional scenarios. Experimental results demonstrate that our method significantly outperforms baseline models across all evaluation metrics (BLEU: 0.361, ROUGE: 0.521, METEOR: 0.443, F1-Score: 0.796), achieving improvements ranging from 21.1% to 73.0% over state-of-the-art methods: specifically, a 21.1% improvement over GECToR in BLEU, 26.5% over GECToR in ROUGE, 30.3% over Deep Edit in METEOR, and 11.8% over Deep Edit in F1-Score, with a reduction of approximately 35% in hallucination rates compared to existing approaches. These improvements provide important technical support for intelligent operation and maintenance of power systems, with implications for improving data quality management, enhancing model reliability in safety-critical applications, and enabling scalable knowledge-guided correction frameworks transferable to other industrial domains requiring high data integrity. Full article

This study draws on socio-technical transition theory to examine how multi-actor dynamics among producers, consumers, and the media within an experimental niche—Korea’s regulatory sandbox—shape policy responsiveness and the regulatory speed of governmental responses to emerging technologies, thereby influencing socio-technical transitions. We construct a longitudinal dataset of 2136 sandbox approvals between 2019 and 2025 and 1374 cases in which related legal or administrative adjustments have been completed. Changes in actor couplings before and after sandbox approval are first assessed using Pearson correlation analysis, while temporal lead–lag relationships are identified via vector autoregression (VAR) and Granger causality tests. Building on these dynamic analyses, the study subsequently investigates the determinants of regulatory response speed using ordered logistic regression, incorporating government policy orientation (progressive vs. conservative) as a moderating variable. The results show, first, that the strong producer–consumer coupling observed prior to sandbox approval weakens afterwards, whereas the consumer–media linkage becomes substantially stronger. Second, the time-series analysis of technologies within the regulatory sandbox reveals a typical technology-push pattern and a self-reinforcing feedback loop. Specifically, producer activity initiates the signal sequence, preceding consumer reactions; subsequently, media coverage significantly drives consumer engagement, and the resulting increase in consumer attention, in turn, stimulates further media coverage. Third, in the ordered logit model, media activity accelerates legal and regulatory reform, whereas consumer activity acts as a delaying factor, with producer activity showing no significant direct effect. Finally, government policy orientation systematically moderates the magnitude and direction of these effects. Overall, the study proposes an actor-centered mechanism in which learning generated in the sandbox is externalized through consumer–media channels and translated into regulatory pacing. Based on these findings, we derive practical implications for firms and regulators regarding proactive media engagement, transparent use of evidence, institutionalized channels for consumer input, and robust feedback standards that support sustainable commercialization of emerging technologies. Full article

Many data processing applications involve binary matrices for storing digital information. At present, there are limited results in the literature about algorithms for inverting large binary matrices. This paper contributes the following three results. First, the divide-and-conquer methods for efficiently inverting large matrices over finite fields such as Strassen’s matrix inversion often fail on singular sub-blocks, even if the original matrix is non-singular. It is proposed to combine Strassen’s method with the PLU factorization at each recursive step in order to obtain robust pivoting, which correctly inverts all non-singular matrices over any finite field. The resulting algorithm is shown to maintain the sub-cubic time complexity. Second, although there are theoretical studies on how to systematically enumerate all invertible matrices over finite fields without redundancy, no practical algorithm has been reported in the literature that is easy to understand and also suitable for enumerating large matrices. The use of Bruhat decomposition has been proposed to enumerate all invertible matrices. It leverages the linear group-theoretic structure and defines an ordered sequence of invertible matrices, so that each matrix is generated exactly once. Third, large binary matrices have about 29% probability to be invertible. In some applications, it may be desirable to repair the singular matrices by performing a small number of bit-flips. It is shown that the minimum number of bit-flips is equal to the matrix rank deficiency, i.e., the minimum Hamming distance from the general linear group. The required bit-flips are identified by pivoting during the matrix inversion, so the matrix rank can be restored. The correctness and the time complexity of the proposed algorithms were verified both theoretically and empirically. The reference implementation of these algorithms in C++ is available on Github. Full article

Gastrointestinal stromal tumors (GISTs) are the most common mesenchymal tumors of the gastrointestinal tract. While imatinib revolutionized first-line therapy, resistance and specific mutation profiles necessitate subsequent generations of tyrosine kinase inhibitors (TKIs). Sunitinib, regorafenib, and avapritinib represent second-line, third-line, and mutation-specific therapies, respectively, offering improved precision and disease control. This review summarizes clinical trial evidence, real-world data, and translational studies evaluating the efficacy, safety, and mechanistic basis of second- and third-generation TKIs in GIST. Emphasis is placed on therapeutic sequencing, resistance mechanisms, and molecularly guided treatment selection. Sunitinib, a multitargeted TKI inhibiting KIT, PDGFR, and VEGFR, provides effective disease control in imatinib-resistant or intolerant patients. Regorafenib, a broad-spectrum multikinase inhibitor, improves progression-free survival in refractory GIST and targets additional angiogenic and oncogenic pathways. Avapritinib, a next-generation TKI, selectively inhibits PDGFRA D842V and KIT exon 17 mutations, addressing a previously untreatable, mutation-driven subgroup. Integration of these agents into treatment algorithms exemplifies a shift toward personalized therapy, with outcomes guided by mutation profiling and biomarker-driven decisions. Second- and third-generation TKIs have transformed the management of advanced GIST, extending survival and offering mutation-specific precision therapy. Ongoing research into resistance mechanisms, combination strategies, and novel inhibitors promises further optimization of patient-centered care. Full article

ESKAPE bacteria are a major global threat due to their rapid antibiotic resistance acquisition and severe healthcare-associated infections. Effective countermeasures require epidemiological surveillance and resistance transmission studies, particularly for antimicrobial-resistant (AMR) colonization in intensive care unit (ICU) patients. Whole-genome sequencing (WGS) provides critical information on resistance spread and mechanisms. In the provided protocol, rectal and oropharyngeal swabs, or endotracheal aspirate/bronchoalveolar lavage for intubated patients, are collected at ICU admission and twice weekly. Patient interviews and medical records identify risk factors for resistant microflora. Samples undergo cultivation, species identification, antibiotic susceptibility testing, and DNA extraction. Sequencing is performed using second- and third-generation platforms, with selected isolates subject to hybrid genome assembly. Resistance genes, virulence factors, and typing profiles (MLST, cgMLST) are determined. This protocol characterizes the ICU patient colonization by AMR pathogens, including species distribution, phenotypic and genotypic resistance profiles, clonal structure, and temporal changes. It estimates detection frequency and colonization patterns at each locus, identifies key risk factors, including prior community or inter-facility exposure, and analyzes associations between risk factors and admission colonization. The study aims to estimate AMR infection risk and severity in ICU patients through the comprehensive analysis of colonization dynamics, resistance patterns, and clonal characteristics using WGS data on pathogen composition and AMR trends. Full article

Course recommendation is a critical service in Intelligent Tutoring Systems (ITS) that helps learners discover relevant courses from massive online educational platforms. Despite substantial progress in this field, two key challenges remain unresolved: (1) existing methods fail to leverage the differences in learners’ interests across different courses during knowledge propagation processes, and (2) while sequential relationships have been considered in course recommendations, there is still significant room for improvement in effectively integrating sequential patterns with knowledge-graph-based approaches. To overcome these limitations, we propose PGDB (Preference-aware Graph Diffusion network and Bi-LSTM), an innovative end-to-end framework for course recommendation. Our model consists of four key components: First, a course knowledge graph diffusion module recursively collects multiple knowledge triples related to learners to construct their knowledge background. Second, a preference-aware diffusion attention mechanism analyzes learners’ preferences for courses and relational paths using multi-head attention, effectively distinguishing semantic diversity across different contexts and capturing varying learner interests during knowledge transmission. Third, a temporal sequence modeling module utilizes bidirectional long short-term memory networks to identify learners’ interest evolution patterns, generating learner-dependent representations that efficiently leverage sequential relationships between courses. Finally, a prediction module combines the final representations of learners and courses to output selection probabilities for candidate courses. Extensive experimental results demonstrate that PGDB significantly outperforms state-of-the-art baseline models across multiple evaluation metrics, validating the effectiveness of our approach in addressing data sparsity and sequential modeling challenges in course recommendation systems. Full article

Post-5G and 6G telecommunication infrastructures face critical information security challenges due to increasing network complexity and sophisticated cyberattacks. Traditional intrusion detection systems based on statistical traffic analysis struggle to identify advanced threats that exploit semantic-level vulnerabilities in modern communication protocols. This paper proposes a Transformer-based intrusion detection system specifically designed for post-5G and 6G networks. Our approach integrates three key innovations: First, a comprehensive feature extraction method capturing both semantic content characteristics and communication behavior patterns. Second, a dynamic semantic embedding mechanism that adaptively adjusts positional encoding based on semantic context changes. Third, a Transformer-based classifier with multi-head attention mechanisms to model long-range dependencies in attack sequences. Extensive experiments on CICIDS2017 and UNSW-NB15 datasets demonstrate superior performance compared to LSTM, GRU, and CNN baselines across multiple evaluation metrics. Robustness testing and cross-dataset validation confirm strong generalization capability, making the system suitable for deployment in heterogeneous post-5G and 6G telecommunication environments. Full article

Sandy beach resilience faces growing threats from extreme events and intensified human activity. Deep Learning (DL) has emerged as a powerful tool in coastal research, offering strengths in spatial feature extraction, nonlinear sequence modeling, acceleration of physical processes, and integration of multi-source data. This review frames resilience in three technical dimensions—resistance, recovery, and adaptation—and examines DL applications across three domains: first, monitoring and forecasting external forcing, including typhoon tracks and storm surge peak values; second, modeling and simulating beach processes, from rapid hydrodynamic forecasting to medium- and long-term shoreline evolution, and high-resolution sediment transport forecasting; and third, management and decision support, where DL methods and multi-scenario generation expand governance options, and interpretable features with uncertainty quantification enhance risk communication and policy adoption. DL complements traditional models by shortening the “observation–model–decision” cycle, expanding scenario analysis, and improving governance transparency. Challenges remain in cross-domain generalization, robustness in extreme scenarios, and data governance. This review confirms DL’s potential as a technology stack for enhancing sandy beach resilience and provides a methodological foundation for future research. Full article