Search Results (640)

To investigate the structural characteristics of Ir_n clusters (n = 9–30) and their interaction with NH₃, the CALYPSO structure-prediction method was employed to identify the lowest-energy configurations. The Lennard–Jones potential was then used to compute the binding energy and average binding energy, thereby evaluating size-dependent stability. The results show that Ir_n clusters evolve from relatively open motifs to compact three-dimensional frameworks as n increases. Meanwhile, the average binding energy increases overall and exhibits several locally stable size regions, indicating a pronounced size effect. Based on slab and cluster models, NH₃ adsorption was further examined on the Ir₁₃ cluster as a representative system due to its high structural stability as a “magic-number” cluster. The calculated adsorption energies demonstrate that the Ir₁₃ cluster exhibits substantially stronger adsorption than the bulk Ir surface, with low-coordinated Ir atoms playing a key role in strengthening the interaction and enhancing adsorption activity. Adsorption-configuration analysis indicates that NH₃ preferentially binds to active surface sites via the N lone pair. These findings clarify the relationship between structural stability and adsorption performance of Ir clusters and provide theoretical support for Ir-based materials in NH₃ catalytic conversion and high-sensitivity gas detection, and offer insights relevant to improving NH₃ monitoring in underground coal mine environments. Full article

To address the problem of renewable energy curtailment and the need for operational economic optimization in integrated energy systems with high penetration of wind and solar power, a coordinated optimization method integrating thermoelectric decoupling, ammonia-blended combustion technology, and energy storage capacity leasing is proposed. First, a chaotic-improved Latin Hypercube Sampling (C-LHS) method, combined with an improved K-means clustering algorithm, is employed to generate representative wind–solar–load scenarios. This approach improves the efficiency of uncertainty scenario generation while reducing computational burden and maintaining solution accuracy. Secondly, by coordinating the operation of thermal energy storage and electric boilers, the “heat-led power generation” constraint is relaxed, and, in combination with ammonia-blended combustion in combined heat and power (CHP) units, the system’s flexibility and renewable energy accommodation capability are enhanced. Finally, with the objective of minimizing total operating cost, a day-ahead scheduling model incorporating electrical energy storage (EES) leasing optimization is established. For EES, under a shared energy storage market mechanism, the golden section search (GSS) algorithm is employed to optimize the day-ahead leasing capacity. The simulation results demonstrate that the proposed method improves renewable energy accommodation while maintaining economic performance, and effectively reduces the overall operating cost of the system. These findings confirm the effectiveness of the proposed strategy in enhancing both system flexibility and economic performance. Full article

Lightweight day-ahead photovoltaic (PV) forecasting models encounter a significant technical challenge: under resource-constrained deployment conditions, it is difficult to simultaneously address weather-regime heterogeneity, maintain model interpretability, and preserve adaptability as operating conditions evolve. To address this issue, we propose a multimodal short-term photovoltaic (PV) forecasting method that integrates weather-mode partitioning using the Clustering Using Representatives (CURE) algorithm with a self-updating Batch-Lasso model. First, the meteorological-PV dataset is partitioned along two dimensions by combining seasonal grouping with CURE clustering within each season, producing representative weather modes and enhancing the fidelity of weather pattern classification. Second, to extract informative predictors from high-dimensional meteorological inputs while maintaining interpretability, we formulate per-mode Lasso regression and adopt the Fast Iterative Shrinkage-Thresholding Algorithm (FISTA) to efficiently solve for the sparse regression coefficients. Third, we introduce a batch-based self-update and correction mechanism with rollback verification, enabling the mode-specific models to be refreshed as new historical data become available while preventing performance degradation. Compared with representative machine learning baselines, the proposed method maintains competitive accuracy with substantially lower computational and storage overhead, enabling high-frequency and energy-efficient inference on resource-constrained terminals, thereby reducing operational burdens and computational energy costs and better meeting the deployment needs of sustainable energy systems under heterogeneous weather conditions. Full article

The rapid growth of Internet of Things (IoT) and Industrial IoT applications has intensified the demand for low-latency and reliable computation support for deadline-constrained periodic real-time tasks. While unmanned aerial vehicles (UAVs) enabling mobile edge computing (MEC) can reduce latency by bringing compute closer to data sources, terrestrial MEC deployments often suffer from limited coverage and poor adaptability to spatially heterogeneous demand. In this paper, we study a multiple-UAV-assisted MEC system serving cluster-based IoT networks, where cluster heads generate deadline-constrained periodic tasks for offloading under strict deadlines. To ensure practical feasibility in dense urban environments, we benchmark UAV mobility using a realistic 3D collision avoidance navigation graph with shortest-path execution, rather than assuming unconstrained continuous UAV motion in free space. On top of this benchmark, we systematically compare three multi-agent reinforcement learning (MARL) paradigms for joint navigation and periodic task offloading: (i) continuous 3D control MARL that outputs motion commands directly; (ii) discrete graph-based MARL that selects collision-free shortest paths; and (iii) asynchronous macro-action MARL. Using a high-fidelity 3D digital twin of San Francisco, we evaluate these paradigms under a unified protocol in terms of offloading success, end-to-end latency, and energy consumption. The results reveal clear performance trade-offs induced by realistic 3D collision avoidance constraints and provide actionable insights for designing UAV-assisted MEC systems supporting periodic real-time task offloading. Full article

Acute myeloid leukemia (AML) is an aggressive hematologic malignancy characterized by profound molecular heterogeneity and high relapse rates, posing significant clinical challenges. Programmed cell death (PCD), encompassing diverse regulated modalities such as apoptosis, necroptosis, and ferroptosis, plays a key role in leukemogenesis and therapeutic response; however, a comprehensive prognostic framework integrating multi-modal PCD pathways in AML remains elusive. In this study, we performed a systematic transcriptomic analysis of 1624 genes associated with 13 distinct PCD forms. A novel computational pipeline combining a variational autoencoder (VAE) for dimensionality reduction and a multilayer perceptron (MLP) for classification was employed to identify robust PCD-related biomarkers, interpreted via SHapley Additive exPlanations (SHAP) analysis. This approach identified 48 candidate genes with discriminative potential between AML and normal bone marrow. Unsupervised consensus clustering based on these genes delineated two molecular subtypes exhibiting divergent clinical outcomes and immune microenvironment profiles. The subtype demonstrated an immunosuppressive phenotype, characterized by enriched regulatory T cells, M2 macrophages, and elevated expression of inhibitory immune checkpoints, correlating with inferior survival. We developed an 8-gene prognostic signature (SORL1, PIK3R5, RIPK3, ELANE, GPX1, VNN1, CD74, and IL3RA) that effectively categorized patients into high- and low-risk groups with notable survival differences, validated across independent cohorts. A prognostic nomogram combining the risk score, age, and cytogenetic risk enhanced the prediction accuracy for overall survival. Our study presents an integrative model that connects multi-modal PCD pathways to AML prognosis, offering a new molecular subtyping system and a clinically applicable risk assessment tool for improved prognostication and personalized treatment strategies. Full article

The rapid expansion of e-commerce and on-demand logistics has intensified the need for cost-effective and reliable urban distribution systems. This paper investigates an energy-aware routing problem for electric vehicle fleets operating from multiple depots under time-varying traffic conditions. We propose a novel multi-depot vehicle routing model that jointly incorporates time-dependent travel speeds, simultaneous pickup and delivery operations, and time window constraints. The model explicitly captures key operational realities, including battery capacity limitations, load- and speed-dependent energy consumption, synchronized pickup-delivery requirements, and soft time windows. The objective is to minimize total operational cost by simultaneously optimizing depot assignments, vehicle routes, and service schedules. Given the NP-hard nature of the problem, we develop a two-stage heuristic solution framework. In the first stage, a spatio-temporal clustering strategy is employed to assign customers to depots efficiently. In the second stage, route construction and improvement are performed using an enhanced Adaptive Large Neighborhood Search (ALNS) algorithm equipped with problem-specific destroy and repair operators. Computational experiments on adapted benchmark instances demonstrate that the proposed approach consistently produces high-quality solutions and exhibits robust convergence behavior. In addition, sensitivity analyses provide managerial insights, revealing an optimal range of vehicle energy capacity and an economically efficient speed band that balances travel time and energy consumption. Full article

Segmenting non-rigid objects such as smoke in video requires effective utilization of temporal information, which remains challenging due to their irregular deformation and complex appearance variations. Based on our previously proposed DeffNet for industrial fumes video segmentation, this letter presents a novel adaptive frame selection algorithm that employs fuzzy logic control to dynamically optimize the temporal processing step size for the specific task of industrial smoke video segmentation. Our method quantifies inter-frame variation using the Structural Similarity Index (SSIM) and Normalized Cross-Correlation (NCC) as inputs to a fuzzy inference system. Gaussian membership functions, shaped via K-means clustering, and a five-rule fuzzy system are designed to determine the optimal step size, maximizing informative dynamic feature extraction while minimizing redundant computation. As a lightweight front-end module, the algorithm integrates seamlessly into the existing DeffNet segmentation framework without reconstructing new network architecture. Extensive experiments on a dedicated industrial smoke video dataset demonstrate that our approach effectively improves the segmentation performance of DeffNet, achieving 84.27% Intersection over Union (IoU) while maintaining a high inference speed of 39.71 FPS. This work provides an efficient and scene-specific solution for temporal modeling in industrial smoke non-rigid object segmentation and offers a practical improved strategy for DeffNet in real-time industrial smoke monitoring. Full article

Robotic manipulators were initially introduced to replace repetitive human labor and have since evolved to perform complex tasks in dynamic environments. In such systems, imitation learning and reinforcement learning models capable of real-time trajectory generation are widely applied. Among these approaches, imitation learning enables rapid training when high-quality datasets are available. However, it suffers from high costs associated with collecting expert demonstration data and significant performance variability depending on data quality. Recently, learning approaches utilizing large-scale datasets have been explored, but they often struggle to guarantee reliable performance in tasks requiring precise control and incur substantial computational costs for model construction, limiting their applicability as a general-purpose learning strategy. To address these limitations, this paper proposes an imitation learning framework that integrates sampling-based motion planning with a hybrid data curation strategy. The proposed method employs a sampling-based planner (e.g., RRT*) to generate diverse physically feasible trajectories, thereby reducing the cost of acquiring expert demonstration data. The generated trajectories are then curated through clustering-based grouping and rule-based filtering to select high-quality training samples from large-scale datasets. The proposed framework automatically generates physically feasible trajectories while selecting high-quality data from large trajectory pools, thereby improving training stability and reducing data-related costs. Experimental results demonstrate that the proposed method achieves an average success rate of 79.1% (95% CI: 74.3–83.2%) and produces trajectories with shorter trajectories, lower final distances, and reduced joint movements compared to conventional filtering methods. Full article

Infected URLs are always regarded as a serious threat to cybersecurity, serving as pathways to phishing, maliciousness, and other offenses. Although transformer-based models have demonstrated good performance in malicious URL detection, their high computational cost and latency make them impractical for deployment in real-time or resource-constrained systems. Allocated on the basis of knowledge distillation (KD), lightweight models tend to be efficient but are commonly not sufficiently discriminative to distinguish between malicious and benign URLs with non-cataclysmic lexical overlaps, particularly when dealing with an imbalanced dataset. In order to address these issues, we propose Contra-KD, a lightweight transformer model that incorporates contrastive learning (CL) and KD. This proposed framework imposes structured embedding matching, allowing the student model to learn more meaningful and generalized depictions. Contra-KD uses a compact 6-layer student transformer architecture based on ELECTRA to scale parameters up and can achieve more than 90% computational fidelity with a high accuracy. In this scheme, CL improves the feature of discrimination by semantically clustering similar URLs and separating different URLs. This tendency serves to limit confusion, especially when a common lexical trait is held between two words and/or in the presence of adversarial obfuscation. Through a large-scale publicly available Kaggle dataset of 651,191 URLs in imbalanced scenarios, the proposed Contra-KD can achieve 99.05% accuracy, 99.96% ROC-AUC, and 98.18% MCC which are superior to their counterparts including lightweight models and transformer-based ones. To summarize, Contra-KD proposes an efficient transformer architecture that is both small and effective in computation while delivering stable detection performance. Full article

Vegetation coverage is a critical indicator for assessing urban ecosystems and is essential for sustainable development. However, the evolution patterns and driving mechanisms of vegetation change at the urban agglomeration scale remain underexplored. This study used the Google Earth Engine (GEE) to compute the kernel Normalized Difference Vegetation Index (kNDVI) for the Beibu Gulf Urban Agglomeration (BGUA), an important emerging coastal urban cluster in southern China, from 2000 to 2022. Trend analysis was employed to examine spatiotemporal changes in kNDVI, and an interpretable machine learning framework was applied to quantify the nonlinear, spatially heterogeneous effects of environmental and anthropogenic drivers. The results show that (1) kNDVI showed a general increasing trend, with medium-to-high kNDVI predominating. Approximately 91.91% of the region maintained an improving trend, whereas vegetation degradation concentrated in the core urban areas. (2) The Categorical Boosting model demonstrated superior performance in predicting kNDVI compared to other machine learning models. (3) The SHAP analysis identified land cover, elevation, and nighttime lights as the primary determinants of kNDVI change. These factors exhibited significant spatial heterogeneity in their nonlinear effects. These findings provide theoretical insights and practical guidance for ecological planning and environmental management in urban agglomerations. Full article

Colorectal cancer (CRC) remains a leading cause of cancer mortality globally, yet current histopathological diagnostics capture only limited features. This study aimed to discover subtle, prognostically significant histomorphological patterns in CRC tissues using unsupervised deep learning. We developed a framework integrating convolutional neural networks with deep clustering, trained on 23,341 image patches from 493 patients. We identified 30 distinct histomorphological clusters from CRC tissue images. Through univariate and multivariate survival analyses, three clusters (Cluster13, Cluster19, and Cluster24) were consistently associated with patient prognosis. These clusters were integrated with clinical factors (T stage, N stage, and differentiation degree) to construct a prognostic risk model. Patients stratified into high-risk and low-risk groups based on model predictions showed significant survival differences in both the training set (N = 493) and an independent validation set (N = 2590). Furthermore, logistic regression and multivariate Cox analyses demonstrated that incorporating the three histomorphological clusters alongside clinical factors yielded a modest but statistically significant improvement in predictive performance compared to clinical factors alone, indicating their complementary value for prognosis. This work demonstrates that computational pathology can uncover novel, visually elusive morphological features with independent prognostic value, offering potential to refine CRC patient stratification and inform clinical decision-making. Full article

Spatial Big Data mining is often hindered by high computational complexity and the intrinsic autocorrelation of georeferenced records. To address these challenges, this study proposes an architectural optimization framework for the CHSMST+ algorithm, designated as CHSMST+MR. Rather than introducing a brand-new clustering paradigm, the framework focuses on a Distributed Spatial Cardinality Reduction (DSCR) layer that aggregates redundant spatial records before the core iterative mining logic begins. By transforming raw records into a weighted key-value representation within the Apache Spark environment, the proposed approach significantly mitigates the shuffling bottleneck common in distributed systems. Experimental validation using high-density biological datasets demonstrates an average execution-time reduction of 51.36%, with performance gains reaching up to 79.96% in specific high-redundancy scenarios. The results, obtained through controlled local emulation, confirm that this architectural optimization provides a scalable, deterministic, and lossless solution for accelerating spatial clustering. This work contributes a methodological path for enhancing the performance of iterative spatial mining algorithms in environments characterized by massive data density and coordinate redundancy. Full article

Thermal comfort and adequate sunlight exposure are essential for maintaining the health of older adults. Although multi-objective optimization (MOO) has been increasingly applied to improve environmental performance in spatial design, most existing studies still rely on computationally expensive physical simulations, and their optimization results often lack interpretability and operability in early design decision-making. To address these issues, this study proposes a collaborative optimization framework that integrates machine learning surrogate models with neural visualization tools to support performance-driven design of age-friendly outdoor spaces at the early stage. Based on survey data from 46 typical Beijing communities, we constructed a parametric model with three objectives: minimizing summer UTCI, maximizing winter UTCI, and maximizing sunlight duration. An XGBoost model is adopted as a surrogate to accelerate performance prediction, while a self-organizing map (SOM) was applied to cluster and visualize Pareto-optimal solutions. The results indicate that the surrogate model achieves high predictive accuracy and reduces overall computational time by approximately 45% compared with conventional physical simulations. Moreover, the SOM-based visual decision process compresses the high-dimensional solution space and reduces candidate schemes by more than 90%, enabling rapid identification of design solutions that balance environmental performance and spatial morphology. The proposed framework improves both computational efficiency and decision support capacity for performance-oriented spatial design and provides a novel methodological reference for the environmental renewal of age-friendly outdoor spaces. Full article

Nonlinear systems with multiple roots arise frequently in biomedical and engineering models, yet their reliable numerical solution remains a challenging task. Many classical methods suffer from sensitivity to initial guesses, reduced convergence rates, and loss of accuracy in the presence of multiple or clustered solutions. In addition, the exploitation of parallelism to improve robustness and computational efficiency has received limited attention. In this work, we propose a high-accuracy parallel numerical framework of fourth-order convergence for the simultaneous approximation of all solutions of nonlinear systems with multiple roots. The proposed scheme is derivative-free and structurally decoupled, enabling efficient parallel implementation and robust convergence even when reliable initial approximations are unavailable. The effectiveness of the method is demonstrated on representative biomedical engineering models, including a glucose–insulin–glucagon regulatory network and a multi-compartment pharmacokinetic system, both exhibiting strong nonlinearity and multistability. Numerical experiments confirm stable convergence toward distinct solution clusters, machine-level accuracy, reduced residual norms, and improved computational performance when compared with existing approaches. These results indicate that the proposed framework provides a reliable and efficient alternative for solving nonlinear systems with multiple roots in complex applied settings. Full article

The morphology and connectivity of subsurface fracture networks are critical factors controlling wellbore stability and hydraulic fracturing efficiency. Accurate characterization of the three-dimensional complexity of fractures holds significant importance for engineering safety and performance enhancement. A novel image segmentation model is established in this study. It enhances the iterative threshold method by incorporating simple linear iterative clustering superpixels, ResNet50, and a Gaussian mixture model. The model first divides complex computed tomography images into numerous superpixel images using simple linear iterative clustering superpixel segmentation. Subsequently, ResNet50 is employed to classify these superpixel images. Based on the classification results, the iterative threshold segmentation method is applied to segment different categories of superpixel images accordingly. Following preliminary image segmentation, Gaussian mixture module is used for denoising the segmented fracture images, resulting in high-precision segmented images. The two-dimensional segmented images are then reconstructed in three-dimensional space, and the three-dimensional distribution characteristics of fractures are analyzed. This study concludes that the new fracture segmentation method enables high-precision extraction of fracture regions. Compared with threshold segmentation, the morphological analysis noise value in the two-dimensional images segmented by the method proposed in this study was reduced from 0.21% to 0.08%. Fracture distribution in three-dimensional space is complex, and areas with larger fracture networks exhibit greater complexity in their three-dimensional distribution. Full article