Search Results (252)

This paper presents a two-stage explainable recommendation system for reciprocal affiliate–seller matching that uses machine learning and data science to handle voluminous data and generate personalized ranking lists for each user. In the first stage, a representation learning model was trained to create dense embeddings for affiliates and sellers, ensuring efficient identification of relevant pairs. In the second stage, a learning-to-rank approach was applied to refine the recommendation list based on user suitability and relevance. Diversity-enhancing reranking (maximal marginal relevance/explicit query aspect diversification) and popularity penalties were also implemented, and their effects on accuracy and provider-side diversity were quantified. Model interpretability techniques were used to identify which features affect a recommendation. The system was evaluated on a fully synthetic dataset that mimics the high-level statistics generated by affiliate platforms, and the results were compared against classical baselines (ALS, Bayesian personalized ranking) and ablated variants of the proposed model. While the reported ranking metrics (e.g., normalized discounted cumulative gain at 10 (NDCG@10)) are close to 1.0 under controlled conditions, potential overfitting, synthetic data limitations, and the need for further validation on real-world datasets are addressed. Attributions based on Shapley additive explanations were computed offline for the ranking model and excluded from the online latency budget, which was dominated by approximate nearest neighbors-based retrieval and listwise ranking. Our work demonstrates that high top-K accuracy, diversity-aware reranking, and post hoc explainability can be integrated within a single recommendation pipeline. While initially validated under synthetic evaluation, the pipeline was further assessed on a public real-world behavioral dataset, highlighting deployment challenges in affiliate–seller platforms and revealing practical constraints related to incomplete metadata. Full article

The rapid spread of climate change misinformation across digital platforms undermines scientific literacy, public trust, and evidence-based policy action. Advances in Natural Language Processing (NLP) and Large Language Models (LLMs) create new opportunities for automating the detection and correction of misleading climate-related narratives. This study presents a multi-stage system that employs state-of-the-art large language models such as Generative Pre-trained Transformer 4 (GPT-4), Large Language Model Meta AI (LLaMA) version 3 (LLaMA-3), and RoBERTa-large (Robustly optimized BERT pretraining approach large) to identify, classify, and generate scientifically grounded corrections for climate misinformation. The system integrates several complementary techniques, including transformer-based text classification, semantic similarity scoring using Sentence-BERT, stance detection, and retrieval-augmented generation (RAG) for evidence-grounded debunking. Misinformation instances are detected through a fine-tuned RoBERTa–Multi-Genre Natural Language Inference (MNLI) classifier (RoBERTa-MNLI), grouped using BERTopic, and verified against curated climate-science knowledge sources using BM25 and dense retrieval via FAISS (Facebook AI Similarity Search). The debunking component employs RAG-enhanced GPT-4 to produce accurate and persuasive counter-messages aligned with authoritative scientific reports such as those from the Intergovernmental Panel on Climate Change (IPCC). A diverse dataset of climate misinformation categories covering denialism, cherry-picking of data, false causation narratives, and misleading comparisons is compiled for evaluation. Benchmarking experiments demonstrate that LLM-based models substantially outperform traditional machine-learning baselines such as Support Vector Machines, Logistic Regression, and Random Forests in precision, contextual understanding, and robustness to linguistic variation. Expert assessment further shows that generated debunking messages exhibit higher clarity, scientific accuracy, and persuasive effectiveness compared to conventional fact-checking text. These results highlight the potential of advanced LLM-driven pipelines to provide scalable, real-time mitigation of climate misinformation while offering guidelines for responsible deployment of AI-assisted debunking systems. Full article

The working environment of fruit-picking robots is highly complex, involving numerous obstacles such as branches. Sampling-based algorithms like Rapidly Exploring Random Trees (RRTs) are faster but suffer from low success rates and poor path quality. Deep reinforcement learning (DRL) has excelled in high-degree-of-freedom (DOF) robot path planning, but typically requires substantial computational resources and long training cycles, which limits its applicability in resource-constrained and large-scale agricultural deployments. However, picking robot agents trained by DRL underperform because of the complexity and dynamics of the picking scenes. We propose a real-time policy reconstruction method based on experience retrieval to augment an agent trained by DRL. The key idea is to optimize the agent’s policy during inference rather than retraining, thereby reducing training cost, energy consumption, and data requirements, which are critical factors for sustainable agricultural robotics. We first use Soft Actor–Critic (SAC) to train the agent with simple picking tasks and less episodes. When faced with complex picking tasks, instead of retraining the agent, we reconstruct its policy by retrieving experience from similar tasks and revising action in real time, which is implemented specifically by real-time action evaluation and rejection sampling. Overall, the agent evolves into an augment agent through policy reconstruction, enabling it to perform much better in complex tasks with narrow passages and dense obstacles than the original agent. We test our method both in simulation and in the real world. Results show that the augment agent outperforms the original agent and sampling-based algorithms such as BIT* and AIT* in terms of success rate (+133.3%) and path quality (+60.4%), demonstrating its potential to support reliable, scalable, and sustainable fruit-picking automation. Full article

Urban surface subsidence is primarily induced by intensive above-ground and underground construction activities and excessive groundwater extraction. Integrating InSAR techniques for safety monitoring of urban subway infrastructure is therefore of great significance for urban safety and sustainable development. However, single-track high-spatial-resolution SAR imagery is insufficient to achieve full coverage over large urban areas, and direct mosaicking of inter-track InSAR results may introduce systematic biases, thereby compromising the continuity and consistency of deformation fields at the regional scale. To address this issue, this study proposes an inter-track InSAR correction and mosaicking approach based on the mean vertical deformation difference within overlapping areas, aiming to mitigate the overall offset between deformation results derived from different tracks and to construct a spatially continuous urban surface deformation field. Based on the fused deformation results, subsidence characteristics along subway lines and in key urban infrastructures were further analyzed. The main urban area and the eastern and western new districts of Zhengzhou, a national central city in China, were selected as the study area. A total of 16 Radarsat-2 SAR scenes acquired from two tracks during 2022–2024, with a spatial resolution of 3 m, were processed using the SBAS-InSAR technique to retrieve surface deformation. The results indicate that the mean deformation rate difference in the overlapping areas between the two SAR tracks is approximately −5.54 mm/a. After applying the difference-constrained correction, the coefficient of determination (R²) between the mosaicked InSAR results and leveling observations increased to 0.739, while the MAE and RMSE decreased to 4.706 and 5.538 mm, respectively, demonstrating good stability in achieving inter-track consistency and continuous regional deformation representation. Analysis of the corrected InSAR results reveals that, during 2022–2024, areas exhibiting uplift and subsidence trends accounted for 37.6% and 62.4% of the study area, respectively, while the proportions of cumulative subsidence and uplift areas were 66.45% and 33.55%. In the main urban area, surface deformation rates are generally stable and predominantly within ±5 mm/a, whereas subsidence rates in the eastern new district are significantly higher than those in the main urban area and the western new district. Along subway lines, deformation rates are mainly within ±5 mm/a, with relatively larger deformation observed only in localized sections of the eastern segment of Line 1. Further analysis of typical zones along the subway corridors shows that densely built areas in the western part of the main urban area remain relatively stable, while building-concentrated areas in the eastern region exhibit a persistent relative subsidence trend. Overall, the results demonstrate that the proposed inter-track InSAR mosaicking method based on the mean deformation difference in overlapping areas can effectively support subsidence monitoring and spatial pattern identification along urban subway lines and key regions under relative calibration conditions, providing reliable remote sensing information for refined urban management and infrastructure risk assessment. Full article

Monitoring construction-phase settlement of estuary-crossing tunnels founded on coastal soft soils is critical for risk management, yet dense in situ measurements are often unavailable along linear corridors. This study uses Sentinel-1A ascending SAR imagery (65 scenes, September 2022–August 2025) to retrieve time-series deformation along the Sanya Estuary Channel tunnel (China) using Permanent Scatterer InSAR (PS-InSAR) and Small Baseline Subset InSAR (SBAS-InSAR). The two approaches reveal a consistent subsidence hotspot at Tunnel Section D (DK0+000–DK0+330), while most of the corridor remains within ±5 mm/a. The line-of-sight deformation rates range from −24 to 17.7 mm/year (PS-InSAR) and −29.9 to 18.7 mm/a (SBAS-InSAR). Time-series analysis at representative points in Section D indicates a maximum cumulative settlement of −75.7 mm and a clear acceleration after May 2023. By integrating the deformation results with geological reports, construction logs and rainfall records, we infer that compressible marine clays and interbedded sand/aquifer zones control the hotspot, whereas excavation/dewatering and rainfall-related groundwater fluctuations further promote consolidation. The results provide a practical basis for subsidence risk screening and monitoring prioritization for estuary-crossing infrastructure in coastal soft-soil settings. From a sustainability perspective, the proposed joint PS–SBAS InSAR framework provides a scalable and cost-effective tool for continuous deformation surveillance, supporting preventive maintenance and risk-informed management of urban underground infrastructure. Full article

This study presents a production-ready AI pipeline for audio content processing, implemented within the Youth Radio platform, which serves as an extension of the European School Radio initiative. The system uses a multi-server architecture: an AI Server that runs batch/offline jobs, orchestrated by Apache Airflow, and two Web Servers that deliver all the Backend as well as the Frontend applications, configured with load balancing and redundancy to ensure high availability and fault tolerance. The implemented AI Pipeline includes tasks such as preprocessing, transcription, audio classification and topic modeling. Processed Podcasts are indexed in a Qdrant vector database to facilitate both dense and sparse retrieval while a recommendation system enriches the user’s experience. We summarize design choices and report system-level metrics and task-level indicators (ASR quality after correction, retrieval effectiveness) to guide similar deployments. Full article

The cotton boll opening rate (BOR) is an important indicator for evaluating the physiological maturation process of cotton and the critical stage of yield formation, and it provides essential guidance for subsequent defoliant application and mechanical harvesting. The investigation of cotton BOR usually relies on manual field surveys, which are time-consuming and destructive, making it difficult to achieve large-scale and efficient monitoring. UAV remote sensing technology has been widely used in crop growth monitoring due to its operational flexibility and high image resolution. However, because of the dense growth of the cotton canopy in UAV remote sensing imagery, the boll opening condition in the lower parts of the canopy cannot be completely observed. In contrast, UAV imagery can effectively monitor cotton leaf chlorophyll content (SPAD) and leaf area index (LAI), both of which undergo continuous changes during the boll opening process. Therefore, this study proposes using SPAD and LAI retrieved from UAV multispectral imagery as physiological intermediary variables to construct an empirical statistical equation and compare it with end-to-end machine learning baselines. Multispectral and ground synchronous data (n = 360) were collected in Baibi Town, Anyang, Henan Province, across four dates (8/28, 9/6, 9/13, 9/24). Twenty-eight commonly used vegetation indices were calculated from multispectral imagery, and Pearson’s correlation analysis was conducted to select indices sensitive to cotton SPAD, LAI, and BOR. Prediction models were constructed using the Random Forest (RF), Gradient Boosting Decision Tree (GBDT), Support Vector Machine (SVM), and Partial Least Squares (PLS) models. The results showed that GBDT achieved the best prediction performance for SPAD (R² = 0.86, RMSE = 1.19), while SVM performed best for LAI (R² = 0.77, RMSE = 0.38). The quadratic polynomial equation constructed using SPAD and LAI achieved R² = 0.807 and RMSE = 0.109 in BOR testing, which was significantly better than the baseline model using vegetation indices to directly regress BOR. The method demonstrated stable performance in spatial mapping of BOR during the boll opening period and showed promising potential for guiding defoliant application and harvest timing. Full article

Background/Objective: The integration of machine learning (ML) and deep learning (DL) has significantly enhanced medical image classification, especially in histopathology, by improving diagnostic accuracy and aiding clinical decision making. However, data privacy concerns and restrictions on sharing patient data limit the development of effective DL models. Federated learning (FL) offers a promising solution by enabling collaborative model training across institutions without exposing sensitive data. This systematic review aims to comprehensively evaluate the current state of FL applications in histopathological image classification by identifying prevailing methodologies, datasets, and performance metrics and highlighting existing challenges and future research directions. Methods: Following PRISMA guidelines, 24 studies published between 2020 and 2025 were analyzed. The literature was retrieved from ScienceDirect, IEEE Xplore, MDPI, Springer Nature Link, PubMed, and arXiv. Eligible studies focused on FL-based deep learning models for histopathology image classification with reported performance metrics. Studies unrelated to FL in histopathology or lacking accessible full texts were excluded. Results: The included studies utilized 10 datasets (8 public, 1 private, and 1 unspecified) and reported classification accuracies ranging from 69.37% to 99.72%. FedAvg was the most commonly used aggregation algorithm (14 studies), followed by FedProx, FedDropoutAvg, and custom approaches. Only two studies reported their FL frameworks (Flower and OpenFL). Frequently employed model architectures included VGG, ResNet, DenseNet, and EfficientNet. Performance was typically evaluated using accuracy, precision, recall, and F1-score. Federated learning demonstrates strong potential for privacy-preserving digital pathology applications. However, key challenges remain, including communication overhead, computational demands, and inconsistent reporting standards. Addressing these issues is essential for broader clinical adoption. Conclusions: Future work should prioritize standardized evaluation protocols, efficient aggregation methods, model personalization, robustness, and interpretability, with validation across multi-institutional clinical environments to fully realize the benefits of FL in histopathological image classification. Full article

Focusing on the ecotone from the Qinghai–Tibetan Plateau to the Loess Plateau (QPtoLP), this study firstly constructs a retrieval model of soil moisture in various depth layers based on multi-source remote sensing data by using the two-source energy balance (TSEB) model and soil–vegetation–atmosphere transfer (SVAT) model. And then, it uncovers how the soil moisture changes across various depths in the root zone and discusses the lagging effect of rainfall. This research indicated that the correlation between the retrieved soil moisture and field-monitored values in various depth layers ranged from 0.720 to 0.8414, demonstrating that it is suitable for the retrieval of soil moisture at various depths in the study area. During the growing season, soil moisture experienced a slight decrease from mid-May to mid-June, followed by a partial recovery in mid-June. After a dry spell in July, the soil moisture reached its lowest point, but surface and deep soil moisture levels rebounded to above 0.2 and 0.1 cm³/cm³, respectively, by mid-August. Spatially, the soil moisture was higher in the southern region, characterized by dense human activities, and lower in the northern region, which is dominated by alpine grasslands. Comparing different depths, the soil moisture at a 0–5 cm depth was generally the highest most of the time, except in July, when the 35–50 cm depth had the highest value. Additionally, the surface soil moisture at a 0–5 cm depth indicated frequent fluctuations at elevations above 4000 m. As the soil depth increases, the rainfall lag effect becomes more pronounced, and the lag effect in the 35–50 cm soil layer is three days. Full article

Background: Anatomopathological reports are largely unstructured, which limits automated data extraction, interoperability, and large-scale research. Manual extraction and standardization are costly and difficult to scale. Objective: We developed and evaluated an automated pipeline for entity extraction and multi-ontology normalization of anatomopathological reports. Methods: A corpus of 560 reports from the Military Hospital of Tunis, Tunisia, was manually annotated for three entity types: sample type, test performed, and finding. The entity extraction utilized BioBERT v1.1, while the normalization combined BioClinicalBERT multi-label classification with retrieval-augmented generation, incorporating both dense and BM25 sparse retrieval over SNOMED CT, LOINC, and ICD-11. The performance was measured using precision, recall, F1-score, and statistical tests. Results: BioBERT achieved high extraction performance (F1: 0.97 for the sample type, 0.98 for the test performed, and 0.93 for the finding; overall 0.963, 95% CI: 0.933–0.982), with low absolute errors. For terminology mapping, the combination of BioClinicalBERT and dense retrieval outperformed the standalone and BM25-based approaches (macro-F1: 0.6159 for SNOMED CT, 0.9294 for LOINC, and 0.7201 for ICD-11). Cohen’s Kappa ranged from 0.7829 to 0.9773, indicating substantial to near-perfect agreement. Conclusions: The pipeline provides robust automated extraction and multi-ontology coding of anatomopathological entities, supporting transformer-based named entity recognition with retrieval-augmented generation. However, given the limitations of this study, multi-institutional validation is needed before clinical deployment. Full article

Accurate six-degree-of-freedom (6-DoF) camera pose estimation is essential for augmented reality, robotics navigation, and indoor mapping. Existing pipelines often depend on detailed floorplans, strict Manhattan-world priors, and dense structural annotations, which lead to failures in ambiguous room layouts where multiple rooms appear in a query image and their boundaries may overlap or be partially occluded. We present Render-Rank-Refine, a two-stage framework operating on coarse semantic meshes without requiring textured models or per-scene fine-tuning. First, panoramas rendered from the mesh enable global retrieval of coarse pose hypotheses. Then, perspective views from the top-k candidates are compared to the query via rotation-invariant circular descriptors, which re-ranks the matches before final translation and rotation refinement. Our method increases camera localization accuracy compared to the state-of-the-art SPVLoc baseline by reducing the translation error by 40.4% and the rotation error by 29.7% in ambiguous layouts, as evaluated on the Zillow Indoor Dataset. In terms of inference throughput, our method achieves 25.8–26.4 QPS, (Queries Per Second) which is significantly faster than other recent comparable methods, while maintaining accuracy comparable to or better than the SPVLoc baseline. These results demonstrate robust, near-real-time indoor localization that overcomes structural ambiguities and heavy geometric assumptions. Full article

Current passive microwave satellites like SMAP still face limitations in observational frequency and responsiveness in regions with frequent cloud cover, dense vegetation, or complex terrain, making it difficult to achieve continuous global monitoring with high spatio-temporal resolution. To enhance global high-frequency monitoring capabilities, this study utilizes global reflectivity data provided by the Tianmu-1 (TM-1) constellation since 2023, combined with multiple auxiliary variables, including NDVI, VWC, precipitation, and elevation, to develop a 9 km resolution soil moisture retrieval model. Several spatial clustering and temporal partitioning strategies are incorporated for systematic evaluation. Additionally, since the publicly available TM-1 L1 reflectivity data does not provide separable polarization channels, this study uses DDM/specular point reflectivity as the primary observable quantity for modeling and mitigates non-soil factor interference by introducing multi-source priors such as NDVI, VWC, precipitation, terrain, and roughness. Unlike SMAP’s “single orbit daily fixed local time” observation mode, TM-1, leveraging multi-constellation and multi-orbit reflection geometry, offers more balanced temporal sampling and availability in cloudy, rainy, and mid-to-high latitude regions. This enables temporal gap filling and rapid event response (such as moisture transitions within hours after precipitation events) during periods of SMAP’s quality masking or intermittent data loss. Results indicate that the model combining LC-cluster with seasonal partitioning delivers the best performance at the cluster level, achieving a correlation coefficient (R) of 0.8155 and an unbiased RMSE (ubRMSE) of 0.0689 cm³/cm³, with a particularly strong performance in barren and shrub ecosystems. Comparisons with SMAP and ISMN datasets show that TM-1 is consistent with mainstream products in trend tracking and systematic error control, providing valuable support for global and high-latitude studies of dynamic hydrothermal processes due to its more balanced mid- and high-latitude orbital coverage. Full article

Generative artificial intelligence (AI) and large language models (LLMs) are reshaping the landscape of intelligent educational systems; however, existing solutions often suffer from unstructured resource organization, limited interpretability, and suboptimal retrieval precision. To address these challenges, this study introduces KA-RAG, a course-oriented question answering (QA) framework that integrates a structured Knowledge Graph (KG) with an Agentic Retrieval-Augmented Generation (Agentic-RAG) workflow. The system incorporates a responsive interface, a unified agent controller (ToolPlanner), a course knowledge graph, and a vector-based retrieval subsystem. By combining symbolic graph reasoning with dense semantic retrieval, the proposed dual-retrieval strategy supports interpretable, context-aware responses to course-related queries. Experiments conducted on a graduate-level Pattern Recognition course demonstrate that KA-RAG achieves a retrieval accuracy of 91.4%, semantic consistency of 87.6%, and an average response latency of 2.8 s. User surveys further reveal significant improvements in learning efficiency and satisfaction. The results validate the feasibility of integrating KG and Agentic-RAG techniques for knowledge-grounded educational applications, offering a practical pathway toward intelligent knowledge organization and interactive learning support. Full article

The definition of the recycling and recyclability rates used today relies too heavily on the mass of the recycled materials, to the disadvantage of less dense materials such as polymers. In order to help with this issue, two indices have been created to, respectively, evaluate the retrievability of materials in their end-of-life stage and their reintegrability in new equipment once they have been recycled. These two indices comprise four indicators each, which are themselves divided into 23 sub-indicators. The six formal mathematical principles of the construction of these entities are presented, along with the formulas used for their calculation. Then, a case study is presented: the data of an equipment from a French EEE and sports and leisure distributor have been collected, and all sub-indicators, indicators and indices have been calculated for this equipment, hence assessing the retrievability and reintegrability of its constitutive materials. In conclusion, the precise nature of the indicators and sub-indicators has allowed us to give eco-design recommendations on different aspects of the design process, such as the choice of materials, the mechanical connections and modularity of the product, and its insertion into the waste treatment chain. Full article

Employee onboarding is a key process in workforce integration but is manual, time-consuming, and departmental. This paper presents OnboardGPT v1.0, an intelligent, scalable conversational AI platform to meet this task with automated and personalized onboarding experience through lightweight neural components. The platform uses a feedforward intent classification model, dense semantic retrieval through cosine similarity, and personalization aware of user profiles to deliver context-sensitive and relevant output. A 500-question proprietary dataset about onboarding and annotated answers was constructed to simulate real enterprise conversations from various roles and departments. The platform was launched with a Flask-based web interface that was not third-party API-dependent and enabled multi-turn dialogue, knowledge base searching, and role-aware task instruction. Experimental evaluation on performance indicators such as task success rate, intent classification accuracy, BLEU score, and user satisfaction in simulation demonstrates the system to be effective in offering coherent and actionable onboarding support. The contribution of this work includes a modular, explainable, and deployable AI pipeline suitable for onboarding automation at the enterprise level and lays the foundation for future extensions that include multilingual support, inclusion of long-term memory, and backend system interoperability. Full article