Search Results (1,228)

We propose an advanced electronic nose based on SE-RelationNet for COPD diagnosis with limited breath samples. The model integrates residual blocks, BiGRU layers, and squeeze–excitation attention mechanisms to enhance feature-extraction efficiency. Experimental results demonstrate exceptional performance with minimal samples: in 4-way 1-shot tasks, the model achieves 85.8% mean accuracy (F1-score = 0.852), scaling to 93.3% accuracy (F1-score = 0.931) with four samples per class. Ablation studies confirm that the 5-layer residual structure and single-hidden-layer BiGRU optimize stability (h_F1-score ≤ 0.011). Compared to SiameseNet and ProtoNet, SE-RelationNet shows superior accuracy (>15% improvement in 1-shot tasks). This technology enables COPD detection with as few as one breath sample, facilitating early intervention to mitigate lung cancer risks in COPD patients. Full article

With the increasing uncertainty introduced by the large-scale integration of renewable energy sources, traditional power dispatching methods face significant challenges, including severe frequency fluctuations, substantial forecasting deviations, and the difficulty of balancing economic efficiency with system stability. To address these issues, a deep learning-based dispatching framework is proposed, which integrates spatiotemporal feature extraction with a stability-aware mechanism. A joint forecasting model is constructed using Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) to handle multi-source inputs, while a reinforcement learning-based stability-aware scheduler is developed to manage dynamic system responses. In addition, an uncertainty modeling mechanism combining Dropout and Bayesian networks is incorporated to enhance dispatch robustness. Experiments conducted on real-world power grid and renewable generation datasets demonstrate that the proposed forecasting module achieves approximately a 2.1% improvement in accuracy compared with Autoformer and reduces Mean Absolute Error (MAE) and Root Mean Square Error (RMSE) by 18.1% and 14.1%, respectively, compared with traditional LSTM models. The achieved Mean Absolute Percentage Error (MAPE) of 5.82% outperforms all baseline models. In terms of scheduling performance, the proposed method reduces the total operating cost by 5.8% relative to Autoformer, decreases the frequency deviation from 0.158 Hz to 0.129 Hz, and increases the Critical Clearing Time (CCT) to 2.74 s, significantly enhancing dynamic system stability. Ablation studies reveal that removing the uncertainty modeling module increases the frequency deviation to 0.153 Hz and raises operational costs by approximately 6.9%, confirming the critical role of this module in maintaining robustness. Furthermore, under diverse load profiles and meteorological disturbances, the proposed method maintains stable forecasting accuracy and scheduling policy outputs, demonstrating strong generalization capabilities. Overall, the proposed approach achieves a well-balanced performance in terms of forecasting precision, system stability, and economic efficiency in power grids with high renewable energy penetration, indicating substantial potential for practical deployment and further research. Full article

Adipose tissue inflammation contributes to obesity-induced insulin resistance. However, increasing evidence shows that high BMI (obesity) is not an accurate predictor of poor metabolic health in individuals. The molecular mechanisms regulating the metabolically activated M1 macrophage phenotype in the adipose tissues leading to insulin resistance remain largely unknown. Although the Janus Kinase (Jak)/signal transducer and activator of transcription 3 (Stat3) signaling in myeloid cells are known to promote the M2 phenotype in tumors, we demonstrate here that the Jak2/Stat3 pathway amplifies M1-mediated adipose tissue inflammation and insulin resistance under metabolic challenges. Ablating Jak2 in the myeloid compartment reduces insulin resistance in obese mice, which is associated with a decrease in infiltration of adipose tissue macrophages (ATMs). We show that the adoptive transfer of Jak2-deficient myeloid cells improves insulin sensitivity in obese mice. Furthermore, the protection of obese mice with myeloid-specific Stat3 deficiency against insulin resistance is also associated with reduced tissue infiltration by macrophages. Jak2/Stat3 in the macrophage is required for the production of pro-inflammatory cytokines that promote M1 macrophage polarization in the adipose tissues of obese mice. Moreover, free fatty acids (FFAs) activate Stat3 in macrophages, leading to the induction of M1 cytokines. Silencing the myeloid cell Stat3 with an in vivo siRNA targeted delivery approach reduces metabolically activated pro-inflammatory ATMs, thereby alleviating obesity-induced insulin resistance. These results demonstrate Jak2/Stat3 in myeloid cells is required for obesity-induced insulin resistance and inflammation. Moreover, targeting Stat3 in myeloid cells may be a novel approach to ameliorate obesity-induced insulin resistance. Full article

To investigate the effect of liquid-phase laser ablation on the hardness of WC-Co cemented carbide, this study performed hardness testing, elemental distribution analysis, and XRD phase analysis. The influence of ablation times on the hardness, elemental distribution, and phase composition of WC-Co cemented carbide was examined, and a model describing the hardness evolution mechanism under liquid-phase laser ablation was proposed. The results demonstrated that the hardness of WC-Co cemented carbide increased with the number of ablations. After 14 ablation times, the maximum hardness reached 2800 HV, representing an increase of 51%–56% compared to the matrix hardness. As the number of ablations increased, the content of ditungsten carbide (W₂C) and tungsten carbide (WC) in the cemented carbide increased, the WC grain size decreased, the dislocation density increased, and the distribution became more uniform. The refinement of WC grains and the elevated dislocation density facilitated stronger intergranular bonding, thereby significantly enhancing the material’s hardness. This study provides theoretical guidance for improving the surface mechanical properties of WC-Co cemented carbide tools through liquid-phase laser ablation. Full article

Accurately predicting how quickly people will adopt electric vehicles (EVs) is vital for planning charging stations, managing supply chains, and shaping climate policy. We present a forecasting model that uses three separate Long Short‑Term Memory (LSTM) branches—one for past EV sales, one for infrastructure and policy signals, and one for economic trends. An attention mechanism first highlights the most important weeks in each branch, then decides which branch matters most at any point in time. Trained end‑to‑end on publicly available data, the model beats traditional statistical methods and newer deep learning baselines while remaining small enough to run efficiently. An ablation study shows that every branch and both attention steps improve accuracy, and that adding policy and economic data helps more than relying on EV history alone. Because the network is modular and its attention weights are easy to interpret, it can be extended to produce confidence intervals, include physical constraints, or forecast adoption of other clean‑energy technologies. Full article

Traditional algorithms such as Dijkstra and APF rely on complete environmental information for path planning, which results in numerous constraints during modeling. This not only increases the complexity of the algorithms but also reduces the efficiency and reliability of the planning. Swarm intelligence algorithms possess strong data processing and search capabilities, enabling them to efficiently solve path planning problems in different environments and generate approximately optimal paths. However, swarm intelligence algorithms suffer from issues like premature convergence and a tendency to fall into local optima during the search process. Thus, an improved Artificial Bee Colony-Beetle Antennae Search (IABCBAS) algorithm is proposed. Firstly, Tent chaos and non-uniform variation are introduced into the bee algorithm to enhance population diversity and spatial searchability. Secondly, the stochastic reverse learning mechanism and greedy strategy are incorporated into the beetle antennae search algorithm to improve direction-finding ability and the capacity to escape local optima, respectively. Finally, the weights of the two algorithms are adaptively adjusted to balance global search and local refinement. Results of experiments using nine benchmark functions and four comparative algorithms show that the improved algorithm exhibits superior path point search performance and high stability in both high- and low-dimensional environments, as well as in unimodal and multimodal environments. Ablation experiment results indicate that the optimization strategies introduced in the algorithm effectively improve convergence accuracy and speed during path planning. Results of the path planning experiments show that compared with the comparison algorithms, the average path planning distance of the improved algorithm is reduced by 23.83% in the 2D multi-obstacle environment, and the average planning time is shortened by 27.97% in the 3D surface environment. The improvement in path planning efficiency makes this algorithm of certain value in engineering applications. Full article

Financial market forecasting remains challenging due to complex nonlinear dynamics and regime-dependent behaviors that traditional models struggle to capture effectively. This research introduces the Adaptive Financial Reservoir Network with Hypernetwork Flow (AFRN–HyperFlow) framework, a novel ensemble architecture integrating Echo State Networks, temporal convolutional networks, mixture density networks, adaptive Hypernetworks, and deep state-space models for enhanced financial time-series prediction. Through comprehensive feature engineering incorporating technical indicators, spectral decomposition, reservoir-based representations, and flow dynamics characteristics, the framework achieves superior forecasting performance across diverse market conditions. Experimental validation on 26,817 balanced samples demonstrates exceptional results with an F1-score of 0.8947, representing a 12.3% improvement over State-of-the-Art baseline methods, while maintaining robust performance across asset classes from equities to cryptocurrencies. The adaptive Hypernetwork mechanism enables real-time regime-change detection with 2.3 days average lag and 95% accuracy, while systematic SHAP analysis provides comprehensive interpretability essential for regulatory compliance. Ablation studies reveal Echo State Networks contribute 9.47% performance improvement, validating the architectural design. The AFRN–HyperFlow framework addresses critical limitations in uncertainty quantification, regime adaptability, and interpretability, offering promising directions for next-generation financial forecasting systems incorporating quantum computing and federated learning approaches. Full article

The increasing reliance on Medical Internet of Things (MIoT) devices introduces critical cybersecurity vulnerabilities, necessitating advanced, adaptive defense mechanisms. Recent cyber incidents—such as compromised critical care systems, modified therapeutic device outputs, and fraudulent clinical data inputs—demonstrate that these threats now directly impact life-critical aspects of patient security. In this paper, we introduce a machine learning-enabled Cognitive Cyber-Physical System (ML-CCPS), which is designed to identify and respond to cyber threats in MIoT environments through a layered cognitive architecture. The system is constructed on a feedback-looped architecture integrating hybrid feature modeling, physical behavioral analysis, and Extreme Learning Machine (ELM)-based classification to provide adaptive access control, continuous monitoring, and reliable intrusion detection. ML-CCPS is capable of outperforming benchmark classifiers with an acceptable computational cost, as evidenced by its macro F1-score of 97.8% and an AUC of 99.1% when evaluated with the ToN-IoT dataset. Alongside classification accuracy, the framework has demonstrated reliable behaviour under noisy telemetry, maintained strong efficiency in resource-constrained settings, and scaled effectively with larger numbers of connected devices. Comparative evaluations, radar-style synthesis, and ablation studies further validate its effectiveness in real-time MIoT environments and its ability to detect novel attack types with high reliability. Full article

This paper explores the impact of applying a powder additive in the form of halloysite and mullite on the thermal protection properties of a composite. The authors used CES R70 epoxy resin with CES H72 hardener, modified by varying the amount of powder additive. The composite samples were exposed to a mixture of combustible gases at a temperature of approximately 1000 °C. The primary parameters analyzed during this study were the temperature on the rear surface of the sample and the ablative mass loss of the tested material. The temperature increase on the rear surface of the sample, which was exposed to the hot stream of flammable gases, was measured for 120 s. Another key parameter considered in the data analysis was the ablative mass loss. The charred layer of the sample played a crucial role in this process, as it helped block oxygen diffusion from the boundary layer of the original material. This charred layer absorbed thermal energy until it reached a temperature at which it either oxidized or was mechanically removed due to the erosive effects of the heating factor. The incorporation of mullite reduced the rear surface temperature from 58.9 °C to 49.2 °C, and for halloysite, it was reduced the rear surface temperature to 49.8 °C. The ablative weight loss dropped from 57% to 18.9% for mullite and to 39.9% for halloysite. The speed of mass ablation was reduced from 77.9 mg/s to 25.2 mg/s (mullite) and 52.4 mg/s (halloysite), while the layer thickness loss decreased from 7.4 mm to 2.8 mm (mullite) and 4.4 mm (halloysite). This research is innovative in its use of halloysite and mullite as functional additives to enhance the ablative resistance of polymer composites under extreme thermal conditions. This novel approach not only contributes to a deeper understanding of composite behavior at high temperatures but also opens up new avenues for the development of advanced thermal protection systems. Potential applications of these materials include aerospace structures, fire-resistant components, and protective coatings in environments exposed to intense heat and flame. Full article

Segmentation of skin lesions in dermoscopic images is critical for the accurate diagnosis of skin cancers, particularly malignant melanoma, yet it is hindered by irregular lesion shapes, blurred boundaries, low contrast, and artifacts, such as hair interference. Conventional deep learning methods, typically based on UNet or Transformer architectures, often face limitations in regard to fully exploiting lesion features and incur high computational costs, compromising precise lesion delineation. To overcome these challenges, we propose SGNet, a structure-guided network, integrating a hybrid CNN–Mamba framework for robust skin lesion segmentation. The SGNet employs the Visual Mamba (VMamba) encoder to efficiently extract multi-scale features, followed by the Dual-Domain Boundary Enhancer (DDBE), which refines boundary representations and suppresses noise through spatial and frequency-domain processing. The Semantic-Texture Fusion Unit (STFU) adaptively integrates low-level texture with high-level semantic features, while the Structure-Aware Guidance Module (SAGM) generates coarse segmentation maps to provide global structural guidance. The Guided Multi-Scale Refiner (GMSR) further optimizes boundary details through a multi-scale semantic attention mechanism. Comprehensive experiments based on the ISIC2017, ISIC2018, and PH2 datasets demonstrate SGNet’s superior performance, with average improvements of 3.30% in terms of the mean Intersection over Union (mIoU) value and 1.77% in regard to the Dice Similarity Coefficient (DSC) compared to state-of-the-art methods. Ablation studies confirm the effectiveness of each component, highlighting SGNet’s exceptional accuracy and robust generalization for computer-aided dermatological diagnosis. Full article

Accurate wellbore trajectory prediction is of great significance for enhancing the efficiency and safety of directional drilling in coal mines. However, traditional mechanical analysis methods have high computational complexity, and the existing data-driven models cannot fully integrate non-sequential features such as stratum lithology. To solve these problems, this study proposes a parallel input gated recurrent unit (Pi-GRU) model based on the TensorFlow framework. The GRU network captures the temporal dependencies of sequence data (such as dip angle and azimuth angle), while the BP neural network extracts deep correlations from non-sequence features (such as stratum lithology), thereby achieving multi-source data fusion modeling. Orthogonal experimental design was adopted to optimize the model hyperparameters, and the ablation experiment confirmed the necessity of the parallel architecture. The experimental results obtained based on the data of a certain coal mine in Shanxi Province show that the mean square errors (MSE) of the azimuth and dip angle angles of the Pi-GRU model are 0.06° and 0.01°, respectively. Compared with the emerging CNN-BiLSTM model, they are reduced by 66.67% and 76.92%, respectively. To evaluate the generalization performance of the model, we conducted cross-scenario validation on the dataset of the Dehong Coal Mine. The results showed that even under unknown geological conditions, the Pi-GRU model could still maintain high-precision predictions. The Pi-GRU model not only outperforms existing methods in terms of prediction accuracy, with an inference delay of only 0.21 milliseconds, but also requires much less computing power for training and inference than the maximum computing power of the Jetson TX2 hardware. This proves that the model has good practicability and deployability in the engineering field. It provides a new idea for real-time wellbore trajectory correction in intelligent drilling systems and shows strong application potential in engineering applications. Full article

Accurate heat load forecasting is essential for the efficiency of District Heating Systems (DHS). Still, it is challenged by the need to model long-term temporal dependencies and nonlinear relationships with weather and other factors. This study proposes a hybrid deep learning framework combining an attention-enhanced Neural Basis Expansion Analysis for Time Series (N-BEATS) model and eXtreme Gradient Boosting (XGBoost). The N-BEATS component, with a multi-head self-attention mechanism, captures temporal dynamics, while XGBoost models non-linear impacts of external variables. Predictions are integrated using an optimized weighted averaging strategy. Evaluated on a dataset from 103 heating units, the model outperformed 13 baselines, achieving an MSE of 0.4131, MAE of 0.3732, RMSE of 0.6427, and R² of 0.9664. This corresponds to a reduction of 32.6% in MSE, 32.0% in MAE, and 17.9% in RMSE, and an improvement of 5.1% in R² over the best baseline. Ablation studies and statistical tests confirmed the effectiveness of the attention mechanism and ensemble strategy. This model provides an efficient solution for DHS load forecasting, facilitating optimized energy dispatch and enhancing system performance. Full article

The goal of single-image super-resolution (SISR) tasks is to generate high-definition images from low-quality inputs, with practical uses spanning healthcare diagnostics, aerial imaging, and surveillance systems. Although cnns have considerably improved image reconstruction quality, existing methods still face limitations, including inadequate restoration of high-frequency details, high computational complexity, and insufficient adaptability to complex scenes. To address these challenges, we propose an Omni-dimensional Dynamic Convolutional Network (ODConvNet) tailored for SISR tasks. Specifically, ODConvNet comprises four key components: a Feature Extraction Block (FEB) that captures low-level spatial features; an Omni-dimensional Dynamic Convolution Block (DCB), which utilizes a multidimensional attention mechanism to dynamically reweight convolution kernels across spatial, channel, and kernel dimensions, thereby enhancing feature expressiveness and context modeling; a Deep Feature Extraction Block (DFEB) that stacks multiple convolutional layers with residual connections to progressively extract and fuse high-level features; and a Reconstruction Block (RB) that employs subpixel convolution to upscale features and refine the final HR output. This mechanism significantly enhances feature extraction and effectively captures rich contextual information. Additionally, we employ an improved residual network structure combined with a refined Charbonnier loss function to alleviate gradient vanishing and exploding to enhance the robustness of model training. Extensive experiments conducted on widely used benchmark datasets, including DIV2K, Set5, Set14, B100, and Urban100, demonstrate that, compared with existing deep learning-based SR methods, our ODConvNet method improves Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity Index (SSIM), and the visual quality of SR images is also improved. Ablation studies further validate the effectiveness and contribution of each component in our network. The proposed ODConvNet offers an effective, flexible, and efficient solution for the SISR task and provides promising directions for future research. Full article

The Grasshopper Optimization Algorithm (GOA) has attracted significant attention due to its simplicity and effective search capabilities. However, its performance deteriorates when dealing with high-dimensional or complex optimization tasks. To address these limitations, this study proposes an improved variant of GOA, named Outpost Multi-population GOA (OMGOA). OMGOA integrates two novel mechanisms: the Outpost mechanism, which enhances local exploitation by guiding agents towards high-potential regions, and the multi-population enhanced mechanism, which promotes global exploration and maintains population diversity through parallel evolution and controlled information exchange. Comprehensive experiments were conducted to evaluate the effectiveness of OMGOA. Ablation studies were performed to assess the individual contributions of each mechanism, while multi-dimensional testing was used to verify robustness and scalability. Comparative experiments show that OMGOA has better optimization performance compared to other similar algorithms. In addition, OMGOA was successfully applied to a real-world engineering problem—lithology prediction from petrophysical logs—where it achieved competitive classification performance. Full article

Nested named entity recognition (NER) is a task that identifies hierarchically structured entities, where one entity can contain other entities within its span. This study introduces a nested NER model for few-shot learning environments, addressing the difficulty of building extensive datasets for general named entities. We enhance the Biaffine nested NER model by modifying its output layer to incorporate label semantic information through a novel label description embedding (LDE) approach, improving performance with limited training data. Our method replaces the traditional biaffine classifier with a label attention mechanism that leverages comprehensive natural language descriptions of entity types, encoded using BERT to capture rich semantic relationships between labels and input spans. We conducted comprehensive experiments on four benchmark datasets: GENIA (nested NER), ACE 2004 (nested NER), ACE 2005 (nested NER), and CoNLL 2003 English (flat NER). Performance was evaluated across multiple few-shot scenarios (1-shot, 5-shot, 10-shot, and 20-shot) using F1-measure as the primary metric, with five different random seeds to ensure robust evaluation. We compared our approach against strong baselines including BERT-LSTM-CRF with nested tags, the original Biaffine model, and recent few-shot NER methods (FewNER, FIT, LPNER, SpanNER). Results demonstrate significant improvements across all few-shot scenarios. On GENIA, our LDE model achieves 45.07% F1 in five-shot learning compared to 30.74% for the baseline Biaffine model (46.4% relative improvement). On ACE 2005, we obtain 44.24% vs. 32.38% F1 in five-shot scenarios (36.6% relative improvement). The model shows consistent gains in 10-shot (57.19% vs. 49.50% on ACE 2005) and 20-shot settings (64.50% vs. 58.21% on ACE 2005). Ablation studies confirm that semantic information from label descriptions is the key factor enabling robust few-shot performance. Transfer learning experiments demonstrate the model’s ability to leverage knowledge from related domains. Our findings suggest that incorporating label semantic information can substantially enhance NER models in low-resource settings, opening new possibilities for applying NER in specialized domains or languages with limited annotated data. Full article