Search Results (2,353)

Overfitting remains a major difficulty in training deep neural networks, especially when attempting to achieve good generalization in complex classification tasks. Standard dropout is often employed to address this issue; however, its uniform random inactivation of neurons typically leads to instability and insufficient performance increases. This paper proposes an upgraded regularization technique merging adaptive sigmoidal dropout with weight amplification, seeking to dynamically adjust neuron deactivation depending on weight statistics, activation patterns, and neuron history. The proposed dropout process uses a sigmoid function driven by a temperature parameter to determine deactivation likelihood and incorporates a “neuron recovery” step to restore important activations. Simultaneously, the method amplifies high-magnitude weights to select crucial traits during learning. The proposed method is tested on CIFAR-10, and CIFAR-100 datasets using four unique CNN architectures, including deep and residual-based models, to evaluate the approach. Results demonstrate that the suggested technique consistently outperforms both standard dropout and baseline models without dropout, yielding higher validation accuracy and lower, more stable validation loss across all datasets. In particular, it demonstrated superior convergence and generalization performance on challenging datasets such as CIFAR-100. These findings demonstrate the potential of the proposed technique to improve model robustness and training efficiency and provide an alternative in complex classification tasks. Full article

To address the issue of agricultural robot loss of control caused by GNSS signal degradation or loss in complex agricultural environments such as farmland and orchards, this study proposes a neural network-based SLAM/GNSS fusion localization algorithm aiming to enhance the robot’s localization accuracy and stability in weak or GNSS-denied environments. It achieves multi-sensor observed pose coordinate system unification through coordinate system alignment preprocessing, optimizes SLAM poses via outlier filtering and drift correction, and dynamically adjusts the weights of poses from distinct coordinate systems via a neural network according to the GDOP. Experimental results on the robotic platform demonstrate that, compared to the SLAM algorithm without pose optimization, the proposed SLAM/GNSS fusion localization algorithm reduced the whole process average position deviation by 37%. Compared to the fixed-weight fusion localization algorithm, the proposed SLAM/GNSS fusion localization algorithm achieved a 74% reduction in average position deviation during transitional segments with GNSS signal degradation or recovery. These results validate the superior positioning accuracy and stability of the proposed SLAM/GNSS fusion localization algorithm in weak or GNSS-denied environments. Orchard experimental results demonstrate that, at an average speed of 0.55 m/s, the proposed SLAM/GNSS fusion localization algorithm achieves an overall average position deviation of 0.12 m, with average position deviation of 0.06 m in high GNSS signal quality zones, 0.11 m in transitional sections under signal degradation or recovery, and 0.14 m in fully GNSS-denied environments. These results validate that the proposed SLAM/GNSS fusion localization algorithm maintains high localization accuracy and stability even under conditions of low and highly fluctuating GNSS signal quality, meeting the operational requirements of most agricultural robots. Full article

This paper presents the flow field reconstruction and prediction of powder fuel transport systems based on representative feature extraction from scattering images using deep learning techniques. A laboratory-built powder fuel supply system was used to conduct scattering spectroscopy experiments on boron-based fuel under various flow rate conditions. Based on the acquired scattering images, a prediction and reconstruction method was developed using a deep network framework composed of a Stacked Autoencoder (SAE), a Backpropagation Neural Network (BP), and a Long Short-Term Memory (LSTM) model. The proposed framework enables accurate classification and prediction of the dynamic evolution of flow structures based on learned representations from scattering images. Experimental results show that the feature vectors extracted by the SAE form clearly separable clusters in the latent space, leading to high classification accuracy under varying flow conditions. In the prediction task, the feature vectors predicted by the LSTM exhibit strong agreement with ground truth, with average mean square error, mean absolute error, and r-square values of 0.0027, 0.0398, and 0.9897, respectively. Furthermore, the reconstructed images offer a visual representation of the changing flow field, validating the model’s effectiveness in structure-level recovery. These results suggest that the proposed method provides reliable support for future real-time prediction of powder fuel mass flow rates based on optical sensing and imaging techniques. Full article

This study aimed to assess the extent of vitamin B₁ and B₆ vitamer loss during a single peritoneal dialysis (PD) session using a combination of chromatographic techniques and chemometric analysis. Dialysis effluent samples were collected from 41 PD patients (22 on continuous ambulatory peritoneal dialysis (CAPD) and 19 on automated peritoneal dialysis (APD)) during a standardised peritoneal equilibration test. Concentrations of thiamine monophosphate, thiamine diphosphate (ThDP), pyridoxine, pyridoxal (PL), and pyridoxamine were determined using high-performance liquid chromatography with a fluorescence detector. The analytical method was validated in terms of sensitivity, linearity, accuracy, and recovery. Multiple regression analysis was employed to identify potential clinical and demographic predictors of vitamin washout. All vitamers except pyridoxal 5-phosphate (PLP) were detectable in dialysis effluents. ThDP exhibited the greatest loss among the B₁ forms (ca. 0.05–0.57 mg/24 h), while PL exhibited the most significant loss among the B₆ forms (ca. 0.01–0.19 mg/24 h). Vitamin losses varied depending on the dialysis modality (continuous ambulatory peritoneal dialysis, or CAPD, versus automated peritoneal dialysis, or APD) and the peritoneal transport category. Regression analysis identified body weight, haemoglobin, and haematocrit as independent predictors of ThDP washout (R² = 0.58). No statistically robust models were established for the other vitamers. Even short medical procedures (such as single PD) can result in measurable losses of water-soluble vitamins, particularly ThDP and PL. The results emphasise the importance of personalised vitamin supplementation for PD patients and suggest that body composition and haematological parameters significantly influence the loss of thiamine. Full article

It is critical to provide life-sustaining treatment to OHCA patients before ambulance care arrives. However, incorrectly performed resuscitation maneuvers reduce the chances of survival and recovery for the victims. Therefore, we must train regularly and learn how to do it correctly. To facilitate regular chest compression training, this study aims to improve the accuracy of a chest compression evaluation system using posture estimation and to develop a web application. To analyze and enhance accuracy, the YOLOv8 posture estimation was used to examine compression depth, recoil, and tempo, and its accuracy was compared to that of the manikin, which has evaluation systems. We conducted comparative experiments with different camera angles and heights to optimize the accuracy of the evaluation. The experimental results showed that an angle of 30 degrees and a height of 50 cm produced superior accuracy. For web application development, a system has been designed to allow users to upload videos for analysis and obtain appropriate compression parameters. The usability evaluation of the application confirmed its ease of use and accessibility, and positive feedback was obtained. In the conclusion, these findings suggest that optimizing recording conditions significantly improves the accuracy of posture-based chest compression evaluation. Future work will focus on enhancing real-time feedback functionality and improving the user interface of the web application. Full article

Total Knee Arthroplasty (TKA) is a well-established surgical intervention for the management of end-stage knee osteoarthritis. While the procedure is generally successful, postoperative rehabilitation remains a key determinant of long-term functional outcomes. Traditional rehabilitation protocols, particularly those requiring in-person clinical visits, often encounter limitations in accessibility, patient adherence, and personalization. In response, emerging sensor technologies have introduced innovative solutions to support and enhance recovery following TKA. This review provides a thematically organized synthesis of the current landscape and future directions of sensor-assisted rehabilitation in TKA. It examines four main categories of technologies: wearable sensors (e.g., IMUs, accelerometers, gyroscopes), smart implants, pressure-sensing systems, and mobile health (mHealth) platforms such as ReHub^® and BPMpathway. Evidence from recent randomized controlled trials and systematic reviews demonstrates their effectiveness in tracking mobility, monitoring range of motion (ROM), detecting gait anomalies, and delivering real-time feedback to both patients and clinicians. Despite these advances, several challenges persist, including measurement accuracy in unsupervised environments, the complexity of clinical data integration, and digital literacy gaps among older adults. Nevertheless, the integration of artificial intelligence (AI), predictive analytics, and remote rehabilitation tools is driving a shift toward more adaptive and individualized care models. This paper concludes that sensor-enhanced rehabilitation is no longer a future aspiration but an active transition toward a smarter, more accessible, and patient-centered paradigm in recovery after TKA. Full article

Compressed sensing is widely used in modern resource-constrained sensor networks. However, achieving high-quality and robust signal reconstruction under low sampling rates and noise interference remains challenging. Traditional CS methods have limited performance, so many deep learning-based CS models have been proposed. Although these models show strong fitting capabilities, they often lack the ability to handle complex noise in sensor networks, which affects their performance stability. To address these challenges, this paper proposes SBCS-Net. This framework innovatively expands the iterative process of sparse Bayesian compressed sensing using convolutional neural networks and Transformer. The core of SBCS-Net is to optimize key SBL parameters through end-to-end learning. This can adaptively improve signal sparsity and probabilistically process measurement noise, while fully leveraging the powerful feature extraction and global context modeling capabilities of deep learning modules. To comprehensively evaluate its performance, we conduct systematic experiments on multiple public benchmark datasets. These studies include comparisons with various advanced and traditional compressed sensing methods, comprehensive noise robustness tests, ablation studies of key components, computational complexity analysis, and rigorous statistical significance tests. Extensive experimental results consistently show that SBCS-Net outperforms many mainstream methods in both reconstruction accuracy and visual quality. In particular, it exhibits excellent robustness under challenging conditions such as extremely low sampling rates and strong noise. Therefore, SBCS-Net provides an effective solution for high-fidelity, robust signal recovery in sensor networks and related fields. Full article

Computational imaging spectrometers using broad-bandpass filter arrays with distinct transmission functions are promising implementations of miniaturization. The number of filters is limited by the practical factors. Compressed sensing is used to model the system as linear underdetermined equations for hyperspectral imaging. This paper proposes the following method: parallel dictionary reconstruction and fusion for spectral recovery in computational imaging spectrometers. Orthogonal systems are the dictionary candidates for reconstruction. According to observation of ground objects, the dictionaries are selected from the candidates using the criterion of incoherence. Parallel computations are performed with the selected dictionaries, and spectral recovery is achieved by fusion of the computational results. The method is verified by simulating visible-NIR spectral recovery of typical ground objects. The proposed method has a mean square recovery error of ≤1.73 × 10⁻⁴ and recovery accuracy of ≥0.98 and is both more universal and more stable than those of traditional sparse representation methods. Full article

In this paper, we aim to develop an efficient algorithm for the solving Tensor Robust Principal Component Analysis (TRPCA) problem, which focuses on obtaining a low-rank approximation of a tensor by separating sparse and impulse noise. A common approach is to minimize the convex surrogate of the tensor rank by shrinking its singular values. Due to the existence of various definitions of tensor ranks and their corresponding convex surrogates, numerous studies have explored optimal solutions under different formulations. However, many of these approaches suffer from computational inefficiency primarily due to the repeated use of tensor singular value decomposition in each iteration. To address this issue, we propose a novel TRPCA algorithm that introduces a new convex relaxation for the tensor norm and computes low-rank approximation more efficiently. Specifically, we adopt the tensor average rank and tensor nuclear norm, and further relax the tensor nuclear norm into a sum of the tensor Frobenius norms of the factor tensors. By alternating updates of the truncated factor tensors, our algorithm achieves efficient use of computational resources. Experimental results demonstrate that our algorithm achieves significantly faster performance than existing reference methods known for efficient computation while maintaining high accuracy in recovering low-rank tensors for applications such as color image recovery and background subtraction. Full article

Intravenous infusion is an important clinical medical intervention, and its safety is critical to patient recovery. To mitigate the elevated risk of complications (e.g., air embolism) arising from delayed response to infusion endpoints, this paper designs a flexible double pole capacitive (FPB) sensor, which includes a main pole plate, an adaptive pole plate, and a back shielding electrode. The sensor establishes a mapping between residual liquid volume in the infusion bottle and its equivalent capacitance, enabling a non-contact adaptive monitoring system. The system enables precise quantification of residual liquid levels, suppressing baseline drift induced by environmental temperature/humidity fluctuations and container variations via an adaptive algorithm, without requiring manual calibration, and overcomes the limitations of traditional rigid sensors when adapting to curved containers. Experimental results showed that the system achieved an overall sensitivity of 753.5 $\mathrm{f}\mathrm{F}/\mathrm{m}\mathrm{m}$, main pole plate linearity of 1.99%, and adaptive pole plate linearity of 0.53% across different test subjects, linearity of 0.53% across different test subjects, with liquid level resolution accuracy reaching 1 mm. These results validate the system’s ultra-high resolution (1 mm) and robust adaptability. Full article

7,12-Dimethylbenzo[a]anthracene (DMBA-7,12), a highly toxic and environmentally persistent polycyclic aromatic hydrocarbon (PAH), poses significant threats to marine biodiversity and human health due to its bioaccumulation through the food chain. Conventional chromatographic methods, while achieving comparable detection limits, are hindered by the need for expensive instrumentation and prolonged analysis times, rendering them unsuitable for rapid on-site monitoring of DMBA-7,12 in marine environments. Therefore, the development of novel, efficient detection techniques is imperative. In this study, we have successfully developed an electrochemical immunosensor based on a polydopamine (PDA)–chitosan (CTs) composite interface to overcome existing technical limitations. PDA provides a robust scaffold for antibody immobilization due to its strong adhesive properties, while CTs enhances signal amplification and biocompatibility. The synergistic integration of these materials combines the high efficiency of electrochemical detection with the specificity of antigen–antibody recognition, enabling precise qualitative and quantitative analysis of the target analyte through monitoring changes in the electrochemical properties at the electrode surface. By systematically optimizing key experimental parameters, including buffer pH, probe concentration, and antibody loading, we have constructed the first electrochemical immunosensor for detecting DMBA-7,12 in seawater. The sensor achieved a detection limit as low as 0.42 ng/mL. In spiked seawater samples, the recovery rates ranged from 95.53% to 99.44%, with relative standard deviations (RSDs) ≤ 4.6%, demonstrating excellent accuracy and reliability. This innovative approach offers a cost-effective and efficient solution for the in situ rapid monitoring of trace carcinogens in marine environments, potentially advancing the field of marine pollutant detection technologies. Full article

The accurate prediction of static formation temperature (SFT) is essential for ensuring safety and efficiency in ultra-deep well drilling operations. Excessive downhole temperatures (>150 °C) can degrade drilling fluids, damage temperature-sensitive tools, and pose serious operational risks. Conventional methods for SFT determination—including direct measurement, temperature recovery inversion, and artificial intelligence models—are often limited by post-drilling data dependency, insufficient spatial resolution, high computational costs, or a lack of adaptability to complex wellbore geometries. In this study, we propose a new pseudo-3D Kriging interpolation framework that explicitly incorporates real wellbore trajectories to improve the spatial accuracy and applicability of pre-drilling SFT predictions. By systematically optimizing key hyperparameters (θ = [10, 10], lob = [0.1, 0.1], upb = [20, 200]) and applying a grid resolution of 100 × 100, the model demonstrates high predictive fidelity. Validation using over 5.1 million temperature data points from 113 wells in the Shunbei Oilfield reveals a relative error consistently below 5% and spatial interpolation deviations within 5 °C. The proposed approach enables high-resolution, trajectory-integrated SFT forecasting before drilling with practical computational requirements, thereby supporting proactive thermal risk mitigation and significantly enhancing operational decision-making on ultra-deep wells. Full article

Guadua amplexifolia J. Presl is a Neotropical bamboo native to southern Mexico through Central America to Colombia, where it thrives in riparian zones of the San Jorge River basin. Despite its ecological and socio-economic importance, its environmental dynamics and provision of ecosystem services remain poorly understood. This study (1) quantifies spatial and temporal land use/cover changes in the municipality of Montelíbano between 2002 and 2022 and (2) evaluates the ecosystem services that local communities derive from in 2002, 2012, and 2022, and they were classified in QGIS using G. amplexifolia. We applied a supervised classification of Landsat imagery (2002, 2012, 2022) in QGIS, achieving 85% overall accuracy and a Cohen’s Kappa of 0.82 (n = 45 reference points). For the social assessment, we held participatory workshops and conducted semi-structured interviews with artisans, fishers, authorities, and NGO representatives; responses were manually coded to extract key themes. The results show a 12% decline in total vegetated area from 2002 to 2012, followed by an 8% recovery by 2022, with bamboo-dominated stands following a similar pattern. Communities identified raw material provision (87% of mentions), climate regulation (82%), and cultural–recreational benefits (58%) as the most important services provided by G. amplexifolia. This is the first integrated assessment of G. amplexifolia’s landscape dynamics and community-valued services in the San Jorge basin, highlighting its dual function as a renewable resource and a natural safeguard against environmental risks. Our findings offer targeted recommendations for management practices and land use policies to support the species’ conservation and sustainable utilization. Full article

Accurately monitoring terrestrial water storage (TWS) variations is essential due to global climate change and growing water demands. This study investigates TWS changes in Oregon, USA, using Global Navigation Satellite System (GNSS) data from the Nevada Geodetic Laboratory, Gravity Recovery and Climate Experiment (GRACE) level-3 mascon data from the Jet Propulsion Laboratory (JPL), and Noah model data from the Global Land Data Assimilation System (GLDAS) data. The results show that the GNSS inversion offers superior spatial resolution, clearly capturing a water storage gradient from 300 mm in the Cascades to 20 mm in the basin and accurately distinguishing between mountainous and basin areas. However, the GRACE data exhibit blurred spatial variability, with the equivalent water height amplitude ranging from approximately 100 mm to 145 mm across the study area, making it difficult to resolve terrestrial water storage gradients. Moreover, GLDAS exhibits limitations in mountainous regions. The GNSS can provide continuous dynamic monitoring, with results aligning well with seasonal trends seen in GRACE and GLDAS data, although with a 1–2 months phase lag compared to the precipitation data, reflecting hydrological complexity. Future work may incorporate geological constraints, region-specific elastic models, and regularization strategies to improve monitoring accuracy. This study demonstrates the strong potential of GNSS technology for monitoring TWS dynamics and supporting environmental assessment, disaster warning, and water resource management. Full article

Sitagliptin is an orally bioavailable selective DPP4 inhibitor that reduces blood glucose levels without significant increases in hypoglycemia. The aim of this study was to design and validate an innovative, rapid, and highly sensitive LC–MS/MS assay for the precise measurement of sitagliptin concentrations in human plasma. This analytical method, utilizing sitagliptin-d4 as the internal standard, is performed using only 100 μL of plasma and a liquid–liquid extraction procedure based on methyl tert-butyl ether (MTBE). Chromatographic separation is expertly achieved with a Kinetex^® C18 column under isocratic elution, employing a perfect 1:1 blend of 5 mM ammonium acetate (with 0.04% formic acid) and acetonitrile, and maintaining an efficient flow rate of 0.2 mL/min. Detection occurs in positive ionization mode through multiple reaction monitoring, precisely targeting transitions of m/z 408.2 → 193.0 for sitagliptin and 412.2 → 239.1 for the IS. The total runtime of this assay is under 2 min. Comprehensive validation in line with MFDS and FDA criteria demonstrates outstanding linearity (5–1000 ng/mL, r² > 0.998), alongside impressive levels of accuracy, precision, recovery and sample stability. Due to its minimal sample requirement and high-throughput capability, the validated approach is highly appropriate for pharmacokinetic and bioequivalence assessments involving sitagliptin. Full article