Search Results (823)

Direct methods for solving protein crystal structures from X-ray diffraction data provide an essential approach for validating predicted models while avoiding external model bias. Nevertheless, traditional iterative projection algorithms, including the widely used Difference Map (DiffMap), are often limited by modest phase retrieval success rates. To address this limitation, we introduce a novel Hybrid Difference Map (HDM) algorithm that synergistically combines the strengths of DiffMap and the Hybrid Input–Output (HIO) method through six distinct iterative update rules. HDM retains an optimized DiffMap-style relaxation term for fine-grained density modulation in protein regions while adopting HIO’s efficient negative feedback mechanism for enforcing the solvent flatness constraint. Using the transmembrane photosynthetic reaction center 2uxj as a test case, the first HDM formula, HDM-f1, successfully recovered an atomic-resolution structure directly from random phases under a conventional full-resolution phasing scheme, demonstrating the robust phasing capability of the approach. Systematic evaluation across 22 protein crystal structures (resolution 1.5–3.0 Å, solvent content ≥ 60%) revealed that all six HDM variants outperformed DiffMap, achieving 1.8–3.5× higher success rates (average 2.8×), performing on par with or exceeding HIO under a conventional phasing scheme. Further performance gains were achieved by integrating HDM with advanced strategies: resolution weighting and a genetic algorithm-based evolutionary scheme. The genetic evolution strategy boosted the success rate to nearly 100%, halved the median number of iterations required for convergence, and reduced the final phase error to approximately 35${}^{\circ }$ on average across test structures through averaging of multiple solutions. The resulting electron density maps were of high interpretability, enabling automated model building that produced structures with a backbone RMSD of less than 0.5 Å when compared to their PDB-deposited counterparts. Collectively, the HDM algorithm suite offers a robust, efficient, and adaptable framework for direct phasing, particularly for challenging cases where conventional methods struggle. Our implementation supports all space groups providing an accessible tool for the broader structural biology community. Full article

The spaceborne full-polarimetric (FP) synthetic aperture radar (SAR) is an advanced sensor for high-resolution Earth observation. However, FP data acquired by such a system are prone to distortions induced by ionospheric Faraday rotation (FR). From the perspective of exploiting these distortions, this enables the estimation of the ionospheric FR angle (FRA), and consequently the total electron content, across most global regions (including the extensive ocean areas) using spaceborne FP SAR measurements. The accuracy of FRA estimation, however, is highly sensitive to noise interference. This study addresses denoising in FRA retrieval based on the Bickel–Bates estimator, with a specific focus on noise reduction methods built upon the adaptive Goldstein filter (AGF) that was originally designed for radar interferometric processing. For the first time, three signal-to-noise ratio (SNR)-based AGFs suitable for FRA estimation are investigated. A key feature of these filters is that their SNRs are all defined using the amplitude of the Bickel–Bates estimator signal rather than the FRA estimates themselves. Accordingly, these AGFs are applied to the estimator signal instead of the estimated FRAs. Two of the three AGFs are developed by adopting the mathematical forms of SNRs and filter parameters consistent with the existing SNR-based AGFs for interferogram. The third AGF is newly proposed by utilizing more general mathematical forms of SNR and filter parameter that differ from the first two. Specifically, its SNR definition aligns with that widely used in image processing, and its filter parameter is derived as a function of the defined SNR plus an additionally introduced adjustable factor. The three SNR-based AGFs tailored for FRA estimation are tested and evaluated against existing AGF variants and classical image denoising methods using three sets of FP SAR Datasets acquired by the L-band ALOS PALSAR sensor, encompassing an ocean-only scene, a plain land–ocean combined scene, and a more complex land–ocean combined scene. Experimental results demonstrate that all three filters can effectively mitigate noise, with the newly proposed AGF achieving the best performance among all denoising methods included in the comparison. Full article

Large language models (LLM) have achieved remarkable advances in natural-language understanding and content generation, and LLM-based agents demonstrate strong adaptability, flexibility, and robustness in handling complex tasks and enabling automated decision-making. Determining the operating mode of a power system requires repeated adjustments of boundary conditions to address violations. Conventional approaches include expert-driven power flow calculations and optimal power flow methods, the latter of which often lack clear physical interpretability during the iterative optimization process. This study proposes a novel paradigm for automated computation and adjustment of power system operating modes based on LLM-driven multi-agent systems. The approach leverages the reasoning capabilities of LLMs to enhance the adaptability of power flow adjustment strategies, while multi-agent coordination with power flow calculation modules ensures computational accuracy, enabling a natural-language-guided adaptive operational computation and adjustment process. The framework also incorporates retrieval-augmented generation techniques to access external knowledge bases and databases, further improving the agents’ understanding of system operational patterns and the accuracy of decision-making. This method constitutes an exploratory application of LLMs and multi-agent technologies in power system computational analysis, highlighting the considerable potential of LLMs to extend and enhance traditional power system analysis methodologies. Full article

This paper proposes a PI-Dual-STGCN fault diagnosis model based on a SHAP-LLM joint explanation framework to address issues such as the lack of transparency in the diagnostic process of deep learning models and the weak interpretability of diagnostic results. PI-Dual-STGCN enhances the interpretability of graph data by introducing physical constraints and constructs a dual-graph architecture based on physical topology graphs and signal similarity graphs. The experimental results show that the dual-graph complementary architecture enhances diagnostic accuracy to 99.22%. Second, a general-purpose SHAP-LLM explanation framework is designed: Explainable AI (XAI) technology is used to analyze the decision logic of the diagnostic model and generate visual explanations, establishing a hierarchical knowledge base that includes performance metrics, explanation reliability, and fault experience. Retrieval-Augmented Generation (RAG) technology is innovatively combined to integrate model performance and Shapley Additive Explanations (SHAP) reliability assessment through the main report prompt, while the sub-report prompt enables detailed fault analysis and repair decision generation. Finally, experiments demonstrate that this approach avoids the uncertainty of directly using large models for fault diagnosis: we delegate all fault diagnosis tasks and core explainability tasks to more mature deep learning algorithms and XAI technology and only leverage the powerful textual reasoning capabilities of large models to process pre-quantified, fact-based textual information (e.g., model performance metrics, SHAP explanation results). This method enhances diagnostic transparency through XAI-generated visual and quantitative explanations of model decision logic while reducing the risk of large model hallucinations by restricting large models to reasoning over grounded, fact-based textual content rather than direct fault diagnosis, providing verifiable intelligent decision support for industrial fault diagnosis. Full article

Accurate estimation of soil moisture (SM) and vegetation optical depth (VOD) is essential for understanding land–atmosphere interactions, climate dynamics, and ecosystem processes. While passive microwave missions such as SMAP and SMOS provide reliable global SM and VOD products, they are limited by coarse spatial resolution and infrequent revisit times. Global Navigation Satellite System Reflectometry (GNSS-R) observations, particularly from the Cyclone GNSS (CYGNSS) mission, offer an improved spatiotemporal sampling rate. This study presents a deep learning framework based on an artificial neural network (ANN) for the simultaneous retrieval of SM and VOD from CYGNSS observations across the contiguous United States (CONUS). Ancillary input features, including specular point latitude and longitude (for spatial context), CYGNSS reflectivity and incidence angle (for surface signal characterization), total precipitation and soil temperature (for hydrological context), and soil clay content and surface roughness (for soil properties), are used to improve the estimates. Results demonstrate strong agreement between the predicted and reference values (SMAP SM and SMOS VOD), achieving correlation coefficients of R = 0.83 and 0.89 and RMSE values of 0.063 m³/m³ and 0.088 for SM and VOD, respectively. Temporal analyses show that the ANN accurately reproduces both seasonal and daily variations in SMAP SM and SMOS VOD (R ≈ 0.89). Moreover, the predicted SM and VOD maps show strong agreement with the reference SM and VOD maps (R ≈ 0.93). Additionally, ANN-derived VOD demonstrates strong consistency with above-ground biomass (R ≈ 0.77), canopy height (R ≈ 0.95), leaf area index (R = 96), and vegetation water content (R ≈ 0.90). These results demonstrate the generalizability of the approach and its applicability to broader environmental sensing tasks. Full article

Cyberbullying and hate speech increasingly appear in multi-modal social media posts, where images and text are combined in diverse and fast changing ways across platforms. These posts differ in style, vocabulary and layout, and labeled data are sparse and noisy, which makes it difficult to train detectors that are both reliable and deployable under tight computational budgets. Many high performing systems rely on large vision language backbones, full parameter fine tuning, online retrieval or model ensembles, which raises training and inference costs. We present a parameter efficient cross-platform multi-modal transfer learning framework for cyberbullying and hateful content detection. Our framework has three components. First, we perform domain adaptive pretraining of a compact ViLT backbone on in domain image-text corpora. Second, we apply parameter efficient fine tuning that updates only bias terms, a small subset of LayerNorm parameters and the classification head, leaving the inference computation graph unchanged. Third, we use noise aware knowledge distillation from a stronger teacher built from pretrained text and CLIP based image-text encoders, where only high confidence, temperature scaled predictions are used as soft labels during training, and teacher models and any retrieval components are used only offline. We evaluate primarily on Hateful Memes and use IMDB as an auxiliary text only benchmark to show that the deployment aware PEFT + offline-KD recipe can still be applied when other modalities are unavailable. On Hateful Memes, our student updates only 0.11% of parameters and retain about 96% of the AUROC of full fine-tuning. Full article

Sphaerodactylidae play a crucial role in ecosystems, possessing significant ecological, scientific, and conservation value. They contribute to pest control and the maintenance of ecological balance, and also provide abundant materials for research in evolutionary biology and biodiversity. To refine the phylogenetic position of Teratoscincus scincus within the Sphaerodactylidae using mitogenomic data, this study sequenced the complete mitochondrial genome of T. scincus using the Illumina NovaSeq Xplus platform, and subsequently performed assembly, annotation, and analysis. The phylogenetic relationships of T. scincus within the Sphaerodactylidae were analyzed using 13 protein-coding genes (PCGs) from the mitochondrial genome via Bayesian inference (BI) and maximum likelihood (ML) methods. The complete mitochondrial genome of T. scincus is 16,943 bp in length and consists of 13 PCGs, 22 tRNA genes, 2 rRNA genes, and 1 control region (D-loop). The base composition shows a distinct AT preference, with the highest A + T content (56.3%) found in the PCGs region. A phylogenetic tree was constructed based on the amino acid sequences of 13 PCGs from the mitochondrial genomes of nine Sphaerodactylidae species retrieved from GenBank and the newly sequenced T. scincus generated in this study. The results confirm that T. scincus belongs to the genus Teratoscincus within the family Sphaerodactylidae. Phylogenetic analysis reveals that T. scincus and Teratoscincus keyserlingii cluster into a monophyletic group, suggesting a close phylogenetic relationship. Additionally, the phylogenetic tree provides new molecular evidence for understanding the formation mechanism of Sphaerodactylidae diversity. This study not only enriches the mitochondrial genome database of Sphaerodactylidae but also lays an important foundation for subsequent research on the adaptive evolution and conservation biology of T. scincus. Full article

Artificial Intelligence (AI) is increasingly reshaping how cultural heritage institutions design and deliver digital visitor experiences, particularly through adaptive Augmented Reality (AR) applications. However, most existing AR deployments in museums and galleries remain static, rule-based, and insufficiently responsive to visitors’ contextual, behavioral, and emotional diversity. This paper presents ARIA (Augmented Reality for Interpreting Artefacts), a conceptual and architectural framework for AI-supported, adaptive AR experiences in cultural heritage settings. ARIA is designed to address current limitations in personalization, affect-awareness, and ethical governance by integrating multimodal context sensing, lightweight affect recognition, and AI-driven content personalization within a unified system architecture. The framework combines Retrieval-Augmented Generation (RAG) for controlled, knowledge-grounded narrative adaptation, continuous user modeling, and interoperable Digital Asset Management (DAM), while embedding Human-Centered Design (HCD) and Fairness, Accountability, Transparency, and Ethics (FATE) principles at its core. Emphasis is placed on accountable personalization, privacy-preserving data handling, and curatorial oversight of narrative variation. ARIA is positioned as a design-oriented contribution rather than a fully implemented system. Its architecture, data flows, and adaptive logic are articulated through representative museum use-case scenarios and a structured formative validation process including expert walkthrough evaluation and feasibility analysis, providing a foundation for future prototyping and empirical evaluation. The framework aims to support the development of scalable, ethically grounded, and emotionally responsive AR experiences for next-generation digital museology. Full article

Game-based learning has become an increasingly popular educational methodology due to its ability to enhance student interest and engagement. The aim of this study was to analyze the effect of game-based learning on motivation, self-efficacy, and academic performance in Natural Sciences learning. A systematic review and meta-analytic methodology was employed, following PRISMA guidelines. For this purpose, the databases consulted were Web of Science and Scopus, from which a total of 234 documents were retrieved and reduced to 15 studies after rigorously applying the established eligibility criteria. These studies were included in the systematic review and meta-analysis to ensure the validity and relevance of the meta-analytic findings. The meta-analytic results revealed a very strong and highly significant positive effect across all subgroups, benefiting the experimental groups (Z = 6.29; p < 0.00001). In conclusion, the implementation of game-based learning has a positive impact on motivation, self-efficacy, and academic performance in the teaching and learning of Natural Sciences content. Therefore, its incorporation into pedagogical practices represents an opportunity to strengthen student engagement and promote more meaningful learning. Full article

With the advancement of artificial intelligence and educational informatization, there is a growing demand for intelligent teaching assistance systems in universities. Focusing on the university “Algorithms” course in the computer science department, this study develops a multi-terminal collaborative knowledge service system, Course-Oriented Knowledge Service–Based AI Teaching Assistant System (CKS-AITAS), which consists of a PC terminal and a mobile terminal, where the PC terminal integrates functions including knowledge graph, semantic retrieval, intelligent question-answering, and knowledge recommendation. While the mobile terminal enables classroom check-in and teaching interaction, thus forming a closed-loop platform for teaching organization, resource acquisition, and knowledge inquiry. For the document retrieval module, paragraph-level semantic modeling of textbook content is conducted using Word2Vec, combined with approximate nearest neighbor indexing, and this module achieves an MRR@10 of 0.641 and an average query time of 0.128 s, balancing accuracy and efficiency; the intelligent question-answering module, based on a self-built course FAQ dataset, is trained via the BERT model to enable question matching and answer retrieval, achieving an accuracy rate of 86.3% and an average response time of 0.31 s. Overall, CKS-AITAS meets the core teaching needs of the course, provides an AI-empowered solution for university teaching, and boasts promising application prospects in facilitating education sustainability. Full article

A cultural heritage augmented reality system overlays virtual information onto real-world heritage sites, enabling intuitive exploration and interpretation with spatial and temporal contexts. This study presents the design and implementation of a cognitive Mobile Augmented Reality (MAR) system that integrates KCHDM-based ontologies with large language models (LLMs) to facilitate intelligent exploration of urban heritage. While conventional AR guides often rely on static data, our system introduces a Semantic Retrieval-Augmented Generation (RAG) pipeline anchored in a structured knowledge base modeled after the Korean Cultural Heritage Data Model (KCHDM). This architecture enables the LLM to perform dynamic contextual reasoning, transforming heritage data into adaptive question-answering (Q&A) and interactive knowledge-testing quizzes that are precisely grounded in both historical and spatial contexts. The system supports on-site AR exploration and map-based remote exploration to ensure robust usability and precise spatial alignment of virtual content. To deliver a rich, multisensory experience, the system provides multimodal outputs, integrating text, images, models, and audio narration. Furthermore, the integration of a knowledge sharing repository allows users to review and learn from others’ inquires. This ontology-driven LLM-integrated MAR design enhances semantic accuracy and contextual relevance, demonstrating the potential of MAR for socially enriched urban heritage experiences. Full article

The integration of Educational Robotics (ER) into teaching and learning has increased in recent years. While several studies have described the design and use of simulators, no content analysis has systematically documented the features and contribution of simulators in ER. Therefore, a systematic review of eight databases was conducted. From 1200 retrieved articles, 89 met the inclusion criteria. The emerged articles comprised two distinct categories: (a) simulator framework studies—describing tools or platforms (54 articles)—and (b) simulator-based intervention studies—reporting empirical implementations (35 articles). Each article was analyzed by two independent researchers who recorded the design features of simulators, their domain of use, the educational level at which the implementation occurred, intervention characteristics, teacher involvement in the studies, and the skills they tried to promote. Findings showed that simulators were primarily designed for STEM education. Most operated in coding environments, used 3D visualization, and were freely available. Interventions were more frequent at the tertiary level, with fewer at primary and secondary levels. Many empirical studies that used simulators employed small samples in short durations, limiting the generalizability of the findings. Simulator-based practices were mainly linked to programming, problem solving, and computational thinking. Higher-order competences such as collaboration or metacognition were rarely addressed. Finally, most intervention studies reported either no or only moderate teacher involvement. This article aims to be a basis for researchers who study ER implementation and simultaneously serve as a basis for choosing, designing, and adopting ER simulators as teaching tools. Full article

Large language models (LLMs) have rapidly transformed artificial intelligence applications across industries, yet their integration into production systems has unveiled critical security vulnerabilities, chief among them prompt injection attacks. This comprehensive review synthesizes research from 2023 to 2025, analyzing 45 key sources, industry security reports, and documented real-world exploits. We examine the taxonomy of prompt injection techniques, including direct jailbreaking and indirect injection through external content. The rise of AI agent systems and the Model Context Protocol (MCP) has dramatically expanded attack surfaces, introducing vulnerabilities such as tool poisoning and credential theft. We document critical incidents including GitHub Copilot’s CVE-2025-53773 remote code execution vulnerability (CVSS 9.6) and ChatGPT’s Windows license key exposure. Research demonstrates that just five carefully crafted documents can manipulate AI responses 90% of the time through Retrieval-Augmented Generation (RAG) poisoning. We propose PALADIN, a defense-in-depth framework implementing five protective layers. This review provides actionable mitigation strategies based on OWASP Top 10 for LLM Applications 2025, identifies fundamental limitations including the stochastic nature problem and alignment paradox, and proposes research directions for architecturally secure AI systems. Our analysis reveals that prompt injection represents a fundamental architectural vulnerability requiring defense-in-depth approaches rather than singular solutions. Full article

Supercooled liquid water (SLW) in mixed-phase clouds significantly influences precipitation efficiency and aviation safety. However, a comprehensive understanding of its vertical structure has been hampered by a lack of sustained, vertically resolved observations over the North China Plain. This study presents the first systematic analysis of SLW vertical distribution and microphysics in this region, utilizing a year-long dataset (2022) from synergistic ground-based instruments in Beijing. Our retrieval approach integrates Ka-band cloud radar, microwave radiometer, ceilometer, and radiosonde data, combining fuzzy-logic phase classification with a liquid water content inversion constrained by column liquid water path. Key findings reveal a distinct bimodal seasonality: SLW primarily occurs at mid-to-upper levels (4–7.5 km) during spring and summer, driven by convective lofting, while winter SLW is confined to lower altitudes (1–2 km) under stable atmospheric conditions. The temperature-dependent occurrence probability of SLW clouds has an annual maximum at −12 °C. The diurnal variation in SLW in summer shows peaks in the afternoon and at night, corresponding to convective cloud activity. Spring, autumn, and winter do not exhibit strong diurnal variations. Retrieved microphysical properties, including liquid water content and droplet effective radius, are consistent with in situ aircraft measurements, validating our methodology. This analysis provides a critical observational benchmark and offers actionable insights for improving cloud microphysics parameterizations in models and optimizing weather modification strategies, such as seeding altitude and timing, in this water-stressed region. Full article

Apple pomace is a significant by-product of the juice processing industry and a rich source of bioactive compounds; however, its potential as a valuable resource is currently largely untapped. In this work, the ultrasound–enzyme-assisted extraction (UEAE) was evaluated as an alternative method for the extraction of phenolic compounds and soluble solids from apple pomace. For this purpose, an optimization study was carried out using a Box–Behnken factorial design combined with the response surface methodology to assess the influence of enzyme/substrate ratio (0–10% v/w), extraction time (1–5 h) and temperature (25–55 °C) on three response variables: total phenolic content, DPPH radical scavenging activity and soluble solids content of the extracts. In addition, the phenolic profile of the extracts was also investigated. According to the model, DPPH radical scavenging activity will record the maximum value (0.69 mmol Trolox/L) for a 10% enzyme/substrate ratio, at 42 °C and 1 h extraction time. Extraction with an enzyme/substrate ratio of 8.5% at 41 °C for 1 h gave the highest retrieval of soluble solids content (4.1%) in the extracts. Based on HPLC results, chlorogenic acid, caffeic acid, rutin, and epicatechin were the predominant polyphenols in the extracts. The results confirmed the great potential of apple pomace as an economical source of bioactive compounds, and UEAE enhanced the recovery of phenolic compounds and soluble solids from this underutilized by-product. Full article