Search Results (178)

The existing deep learning-based Airborne Laser Scanning (ALS) point cloud semantic segmentation methods require a large amount of labeled data for training, which is not always feasible in practice. Insufficient training data may lead to over-fitting. To address this issue, we propose a novel Multi-branch Feature Extractor (MFE) and a three-stage transfer learning strategy that conducts pre-training on multi-source ALS data and transfers the model to another dataset with few samples, thereby improving the model’s generalization ability and reducing the need for manual annotation. The proposed MFE is based on a novel multi-branch architecture integrating Neighborhood Embedding Block (NEB) and Point Transformer Block (PTB); it aims to extract heterogeneous features (e.g., geometric features, reflectance features, and internal structural features) by leveraging the parameters contained in ALS point clouds. To address model transfer, a three-stage strategy was developed: (1) A pre-training subtask was employed to pre-train the proposed MFE if the source domain consisted of multi-source ALS data, overcoming parameter differences. (2) A domain adaptation subtask was employed to align cross-domain feature distributions between source and target domains. (3) An incremental learning subtask was proposed for continuous learning of novel categories in the target domain, avoiding catastrophic forgetting. Experiments conducted on the source domain consisted of DALES and Dublin datasets and the target domain consists of ISPRS benchmark dataset. The experimental results show that the proposed method achieved the highest OA of 85.5% and an average F1 score of 74.0% using only 10% training samples, which means the proposed framework can reduce manual annotation by 90% while keeping competitive classification accuracy. Full article

The structural units with different characteristic scales in gradient nanostructured (GS) 316L stainless steel act synergistically to achieve the matching of strength and plasticity, and the intrinsic plasticity of nanoscale and ultrafine grains is fully demonstrated. The macroscopic stress–strain responses of each material unit in the GS surface layer can be measured directly by tension or compression tests on microspecimens. However, the experimental results based on microspecimens do not reflect either the extraordinary strengthening effect caused by non-uniform deformation or the intrinsic plasticity of nanoscale and ultrafine grains. In this paper, a method for constructing depth-dependent constitutive relationships of GS materials was proposed, which combines strain hardening parameter (hardness) with physics-informed neural networks (PINNs). First, the microhardness distribution on the specimen cross-sections was measured after stretching to different strains, and the hardness–strain–force test data were used to construct the depth-dependent PINNs model for the true strain–hardness relationship (PINNs_εH). Hardness–strain–force test data from specimens with uniform coarse grains were used to pre-train the PINNs model for hardness and true stress (PINNs_Hσ), on the basis of which the depth-dependent PINNs_Hσ model for GS materials was constructed by transfer learning. The PINNs_εσ model, which characterizes the depth-dependent constitutive relationships of GS materials, was then constructed using hardness as an intermediate variable. Finally, the accuracy and validation of the PINNs_εσ model were verified by a three-point flexure test and finite element simulation. The modeling method proposed in this study can be used to determine the position-dependent constitutive relationships of heterogeneous materials. Full article

Accurate time-series forecasting plays a vital role in sensor-driven applications such as energy monitoring, traffic flow prediction, and environmental sensing. While most existing approaches focus on extracting local patterns from historical observations, they often overlook the global temporal information embedded in timestamps. However, this information represents a valuable yet underutilized aspect of sensor-based data that can significantly enhance forecasting performance. In this paper, we propose a novel timestamp-guided knowledge distillation framework (TKDF), which integrates both historical and timestamp information through mutual learning between heterogeneous prediction branches to improve forecasting robustness. The framework comprises two complementary branches: a Backbone Model that captures local dependencies from historical sequences, and a Timestamp Mapper that learns global temporal patterns encoded in timestamp features. To enhance information transfer and reduce representational redundancy, a self-distillation mechanism is introduced within the Timestamp Mapper. Extensive experiments on multiple real-world sensor datasets—covering electricity consumption, traffic flow, and meteorological measurements—demonstrate that the TKDF consistently improves the performance of mainstream forecasting models. Full article

In order to address the limitations of the number of label samples in practical accurate classification scenarios and the problems of overfitting and an insufficient generalization ability caused by Few-Shot Learning (FSL) in hyperspectral image classification (HSIC), this paper designs and implements a neural architecture search (NAS) for a few-shot HSI classification method that combines meta learning. Firstly, a multi-source domain learning framework was constructed to integrate heterogeneous natural images and homogeneous remote sensing images to improve the information breadth of few-sample learning, enabling the final network to enhance its generalization ability under limited labeled samples by learning the similarity between different data sources. Secondly, by constructing precise and robust search spaces and deploying different units at different locations, the classification accuracy and model transfer robustness of the final network can be improved. This method fully utilizes spatial texture information and rich category information of multi-source data and transfers the learned meta knowledge to the optimal architecture for HSIC execution through precise and robust search space design, achieving HSIC tasks with limited samples. Experimental results have shown that our proposed method achieved an overall accuracy (OA) of 98.57%, 78.39%, and 98.74% for classification on the Pavia Center, Indian Pine, and WHU-Hi-LongKou datasets, respectively. It is fully demonstrated that utilizing spatial texture information and rich category information of multi-source data, and through precise and robust search space design, the learned meta knowledge is fully transmitted to the optimal architecture for HSIC, perfectly achieving classification tasks with few-shot samples. Full article

Federated learning (FL) has emerged as a powerful approach for privacy-preserving model training in autonomous vehicle networks, where real-world deployments rely on multiple roadside units (RSUs) serving heterogeneous clients with intermittent connectivity. While most research focuses on single-server or hierarchical cloud-based FL, multi-server FL can alleviate the communication bottlenecks of traditional setups. To this end, we propose an edge-based, multi-server FL (MS-FL) framework that combines performance-driven aggregation at each server—including statistical weighting of peer updates and outlier mitigation—with an application layer handover protocol that preserves model updates when vehicles move between RSU coverage areas. We evaluate MS-FL on both MNIST and GTSRB benchmarks under shard- and Dirichlet-based non-IID splits, comparing it against single-server FL and a two-layer edge-plus-cloud baseline. Over multiple communication rounds, MS-FL with the Statistical Performance-Aware Aggregation method and Dynamic Weighted Averaging Aggregation achieved up to a 20-percentage-point improvement in accuracy and consistent gains in precision, recall, and F1-score (95% confidence), while matching the low latency of edge-only schemes and avoiding the extra model transfer delays of cloud-based aggregation. These results demonstrate that coordinated cooperation among servers based on model quality and seamless handovers can accelerate convergence, mitigate data heterogeneity, and deliver robust, privacy-aware learning in connected vehicle environments. Full article

Skin cancer is one of the most prevalent forms of cancer worldwide, and early and accurate diagnosis critically impacts patient outcomes. Given the sensitive nature of medical data and its fragmented distribution across institutions (data silos), privacy-preserving collaborative learning is essential to enable knowledge-sharing without compromising patient confidentiality. While federated learning (FL) offers a promising solution, existing methods struggle with heterogeneous and missing modalities across institutions, which reduce the diagnostic accuracy. To address these challenges, we propose an effective and flexible Personalized Multimodal Federated Learning framework (PMM-FL), which enables efficient cross-client knowledge transfer while maintaining personalized performance under heterogeneous and incomplete modality conditions. Our study contains three key contributions: (1) A hierarchical aggregation strategy that decouples multi-module aggregation from local deployment via global modular-separated aggregation and local client fine-tuning. Unlike conventional FL (which synchronizes all parameters in each round), our method adopts a frequency-adaptive synchronization mechanism, updating parameters based on their stability and functional roles. (2) A multimodal fusion approach based on multitask learning, integrating learnable modality imputation and attention-based feature fusion to handle missing modalities. (3) A custom dataset combining multi-year International Skin Imaging Collaboration(ISIC) challenge data (2018–2024) to ensure comprehensive coverage of diverse skin cancer types. We evaluate PMM-FL through diverse experiment settings, demonstrating its effectiveness in heterogeneous and incomplete modality federated learning settings, achieving 92.32% diagnostic accuracy with only a 2% drop in accuracy under 30% modality missingness, with a 32.9% communication overhead decline compared with baseline FL methods. Full article

Federated edge learning (FEEL) is an innovative approach that facilitates collaborative training among numerous distributed edge devices while eliminating the need to transfer sensitive information. However, the practical deployment of FEEL faces significant constraints, owing to the limited and asymmetric computational and communication resources of these devices, along with their energy availability. To this end, we propose a novel asymmetry-tolerant training approach for FEEL, enabled via simultaneous wireless information and power transfer (SWIPT). This framework leverages SWIPT to offer sustainable energy support for devices while enabling them to train models with varying intensities. Given a limited energy budget, we highlight the critical trade-off between heterogeneous local training intensities and the quality of wireless transmission, suggesting that the design of local training and wireless transmission should be closely integrated, rather than treated as separate entities. To elucidate this perspective, we rigorously derive a new explicit upper bound that captures the combined impact of local training accuracy and the mean square error of wireless aggregation on the convergence performance of FEEL. To maximize overall system performance, we formulate two key optimization problems: the first aims to maximize the energy harvesting capability among all devices, while the second addresses the joint learning–communication optimization under the optimal energy harvesting solution. Comprehensive experiments demonstrate that our proposed framework achieves significant performance improvements compared to existing baselines. Full article

The challenges of globalization and urbanization increasingly impact the Historic Urban Landscape (HUL), yet fine-grained and quantitative methods for evaluating HUL remain limited. Adopting a human-centered perspective, this study introduces a novel framework to quantitatively evaluate HUL through the lens of Historical Appearance Integrity (HAI). An evaluation system comprising four key dimensions (building materials, building colors, decorative details, and streetscape morphology) was constructed using the Analytic Hierarchy Process (AHP). An Elo rating system was subsequently applied to quantify the scores of the indicators. A prediction model was developed based on transfer learning and feature fusion to estimate the scores of the indicators. The model achieved accuracies above 93% and loss values below 0.2 for all four indicators. The framework was applied to the Inner Qinhuai Historical Character Area in Nanjing for validation. Results show that the spatial distribution of HAI in the area exhibits significant spatial heterogeneity. On a 0–100 scale, the average HAI scores were 23.17 for primary roads, 27.73 for secondary roads, and 46.93 for branch roads. This study offers a fine-grained, automated approach to evaluate HAI along urban streets and provides a quantitative reference for heritage conservation and urban renewal strategies. Full article

Soil organic carbon (SOC) is a critical indicator of soil health and carbon sequestration potential. Accurate, efficient, and scalable SOC estimation is essential for sustainable orchard management and climate-resilient agriculture. However, traditional visible–near-infrared (Vis–NIR) spectroscopy often suffers from limited chemical specificity and weak adaptability in heterogeneous soil environments. To overcome these limitations, this study develops a five-stage modeling framework that systematically integrates Fourier Transform Infrared (FTIR) spectroscopy with hybrid machine learning techniques for non-destructive SOC prediction in citrus orchard soils. The proposed framework includes (1) FTIR spectral acquisition; (2) a comparative evaluation of nine spectral preprocessing techniques; (3) dimensionality reduction via three representative feature selection algorithms, namely the Successive Projections Algorithm (SPA), Competitive Adaptive Reweighted Sampling (CARS), and Principal Component Analysis (PCA); (4) regression modeling using six machine learning algorithms, namely the Random Forest (RF), Support Vector Regression (SVR), Gray Wolf Optimized SVR (SVR-GWO), Partial Least Squares Regression (PLSR), Principal Component Regression (PCR), and the Back-propagation Neural Network (BPNN); and (5) comprehensive performance assessments and the identification of the optimal modeling pathway. The results showed that second-derivative (SD) preprocessing significantly enhanced the spectral signal-to-noise ratio. Among feature selection methods, the SPA reduced over 300 spectral bands to 10 informative wavelengths, enabling efficient modeling with minimal information loss. The SD + SPA + RF pipeline achieved the highest prediction performance (R² = 0.84, RMSE = 4.67 g/kg, and RPD = 2.51), outperforming the PLSR and BPNN models. This study presents a reproducible and scalable FTIR-based modeling strategy for SOC estimation in orchard soils. Its adaptive preprocessing, effective variable selection, and ensemble learning integration offer a robust solution for real-time, cost-effective, and transferable carbon monitoring, advancing precision soil sensing in orchard ecosystems. Full article

Near-Infrared Spectroscopy (NIRS) analysis technology faces numerous challenges in industrial applications. Firstly, the generalization capability of models is significantly affected by instrumental heterogeneity, environmental interference, and sample diversity. Traditional modeling methods exhibit certain limitations in handling these factors, making it difficult to achieve effective adaptation across different scenarios. Specifically, data distribution shifts and mismatches in multi-scale features hinder the transferability of models across different crop varieties or instruments from different manufacturers. As a result, the large amount of previously accumulated NIRS and reference data cannot be effectively utilized in modeling for new instruments or new varieties, thereby limiting improvements in modeling efficiency and prediction accuracy. To address these limitations, this study proposes a novel transfer learning framework integrating multi-scale network architecture with Balanced Distribution Adaptation (BDA) to enhance cross-instrument compatibility. The key contributions include: (1) RX-Inception multi-scale structure: Combines Xception’s depthwise separable convolution with ResNet’s residual connections to strengthen global–local feature coupling. (2) Squeeze-and-Excitation (SE) attention: Dynamically recalibrates spectral band weights to enhance discriminative feature representation. (3) Systematic evaluation of six transfer strategies: Comparative analysis of their impacts on model adaptation performance. Experimental results on open corn and pharmaceutical datasets demonstrate that BDSER-InceptionNet achieves state-of-the-art performance on primary instruments. Notably, the proposed Method 6 successfully enables NIRS model sharing from primary to secondary instruments, effectively mitigating spectral discrepancies and significantly improving transfer efficacy. Full article

Optimizing ambulance deployment is a critical task in emergency medical services (EMS), as it directly affects patient outcomes and system efficiency. This study proposes a cyber-secure, machine learning-based framework for predicting region-specific ambulance allocation and response times across South Korea. The model integrates heterogeneous datasets—including demographic profiles, transportation indices, medical infrastructure, and dispatch records from 229 EMS centers—and incorporates real-time IoT streams such as traffic flow and geolocation data to enhance temporal responsiveness. Supervised regression algorithms—Random Forest, XGBoost, and LightGBM—were trained on 2061 center-month observations. Among these, Random Forest achieved the best balance of accuracy and interpretability (MSE = 0.05, RMSE = 0.224). Feature importance analysis revealed that monthly patient transfers, dispatch variability, and high-acuity case frequencies were the most influential predictors, underscoring the temporal and contextual complexity of EMS demand. To support policy decisions, a Lasso-based simulation tool was developed, enabling dynamic scenario testing for optimal ambulance counts and dispatch time estimates. The model also incorporates the coefficient of variation (CV) of workload intensity as a performance metric to guide long-term capacity planning and equity assessment. All components operate within a cyber-secure architecture that ensures end-to-end encryption of sensitive EMS and IoT data, maintaining compliance with privacy regulations such as GDPR and HIPAA. By integrating predictive analytics, real-time data, and operational simulation within a secure framework, this study offers a scalable and resilient solution for data-driven EMS resource planning. Full article

Debonding, especially in plastic materials, refers to the separation occurring at the interface within a bonded structure composed of two or more polymeric layers. Due to the great heterogeneity of materials and layering configurations, highly specialized expertise is often required to detect the presence and extent of such defects. This study presents a novel approach that leverages transfer learning techniques to improve the detection of debonding defects across different surface types using PICUS, an acoustic diagnostic device developed at Roma Tre University for the assessment of defects in heritage wall paintings. Our method leverages a pre-trained deep learning model, adapting it to new material conditions. We designed a planar test object embedded with controlled subsurface cavities to simulate the presence of defects of adhesion and air among the layers. This was rigorously evaluated using non-destructive testing using PICUS, augmented by artificial intelligence (AI). A convolutional neural network (CNN), initially trained on this mock-up, was then fine-tuned via transfer learning on a second test object with distinct geometry and material characteristics. This strategic adaptation to varying physical and acoustic properties led to a significant improvement in classification precision of defect class, from 88% to 95%, demonstrating the effectiveness of transfer learning for robust cross-domain defect detection in challenging diagnostic applications. Full article

The massive integration of variable renewable energy sources (RESs) poses the gradual necessity for new power system architectures with wide implementation of distributed battery energy storage systems (BESSs), which support power system stability, energy management, and control. This research presents a methodology and realization of a set of 11 BESS models based on different machine learning methods. The performance of the proposed models is tested using real-life BESS data, after which a comparative evaluation is presented. Based on the results achieved, a valuable discussion and conclusions about the models’ performance are made. This study compares the results of feedforward neural networks (FNNs), a homogeneous ensemble of FNNs, multiple linear regression, multiple linear regression with polynomial features, decision-tree-based models like XGBoost, CatBoost, and LightGBM, and heterogeneous ensembles of decision tree modes in the day-ahead forecasting of an existing real-life BESS in a PV power plant. A Bayesian hyperparameter search is proposed and implemented for all of the included models. Among the main objectives of this study is to propose hyperparameter optimization for the included models, research the optimal training period for the available data, and find the best model from the ones included in the study. Additional objectives are to compare the test results of heterogeneous and homogeneous ensembles, and grid search vs. Bayesian hyperparameter optimizations. Also, as part of the deep learning FNN analysis study, a customized early stopping function is introduced. The results show that the heterogeneous ensemble model with three decision trees and linear regression as main model achieves the highest average R² of 0.792 and the second-best nRMSE of 0.669% using a 30-day training period. CatBoost provides the best results, with an nRMSE of 0.662% for a 30-day training period, and offers competitive results for R²—0.772. This study underscores the significance of model selection and training period optimization for improving battery performance forecasting in energy management systems. The trained models or pipelines in this study could potentially serve as a foundation for transfer learning in future studies. Full article

As systems in industry become increasingly interconnected and sophisticated, the task of fault detection and diagnosis becomes significantly more difficult. Predictive maintenance, in conjunction with sophisticated multimodal learning methods, has been found to be an effective solution for tackling such challenges. Presently, data are collected across numerous sources, ranging from sensors and operational variables to environmental variables, making it vital to combine these heterogeneous data for effective diagnostics. Advanced learning methods like deep learning, transfer learning, and hybrid models are tailored to processing and aggregating such disparate streams of data, thereby leading to higher diagnostic accuracy. This leads to more efficient and reliable predictive maintenance methods. This paper provides a comprehensive review of how various learning methods are applied to fault diagnosis in interconnected systems, particularly in predictive maintenance. It examines different approaches that integrate data across domains, evaluating how each contributes to improved fault detection and enhanced system reliability. Additionally, it addresses emerging research areas, such as real-time fault detection, innovative data fusion processes, and the increasing application in power grids, manufacturing, and the automation sector. This paper serves as a valuable resource for both researchers and practitioners, emphasizing the significant potential of multimodal learning in advancing fault diagnosis and predictive maintenance within increasingly interconnected and complex systems. Full article

Federated learning (FL) enables privacy-preserving collaborative training by iteratively aggregating locally trained model parameters on a central server while keeping raw data decentralized. However, FL faces critical challenges arising from data heterogeneity, model heterogeneity, and excessive communication costs. To address these issues, we propose a communication-efficient federated learning via knowledge distillation and ternary compression framework (FedDT). First, to mitigate the negative impact of data heterogeneity, we pre-train personalized heterogeneous teacher models for each client and employ knowledge distillation to transfer knowledge from teachers to student models, enhancing convergence speed and generalization capability. Second, to resolve model heterogeneity, we utilize the server-initialized global model as a shared student model across clients, where homogeneous student models mask local architectural variations to align feature representations. Finally, to reduce communication overhead, we introduce a two-level compression strategy that quantizes the distilled student model into ternary weight networks layer by layer, substantially decreasing parameter size. Comprehensive evaluations on both MNIST and Cifar10 datasets confirm that FedDT attains 7.85% higher model accuracy and reduces communication overhead by an average of 78% compared to baseline methods. This approach provides a lightweight solution for FL systems, significantly lowering communication costs while maintaining superior performance. Full article