This paper proposes a dynamic gain-adaptive scheme for simultaneously transmitting and reflecting reconfigurable intelligent surface (STAR-RIS) based non-orthogonal multiple access (NOMA) networks (termed as ASRN) with a transmit antenna selection technique at the base station to serve two vehicular users. Unlike conventional fixed-ranking NOMA, ASRN dynamically allocates power and assigns successive interference cancellation order to users based on their channel conditions. Firstly, we derive closed-form expressions for the system outage probability, asymptotic outage behavior, and diversity order, revealing that the proposed scheme achieves a diversity gain scaling with the number of STAR-RIS elements. Then, a greedy algorithm is further proposed to jointly optimize antenna selection and NOMA users' pairing. Secondly, we introduce a key performance metric (termed as the ergodic sum rate (ESR)) and define the multiple averaging ergodic sum rate (MA-ESR) to evaluate the effectiveness of ASRN scheme in delay-tolerant networking approach, when compared to other schemes. Monte Carlo simulations demonstrate that ASRN significantly outperforms dynamic gain-adaptive STAR-RIS OMA conventional STAR-RIS NOMA/OMA, and decode-and-forward relaying schemes, offering superior outage performance, coding gains, ESR performance and MA-ESR performance, particularly in vehicular environments.

Objectives . The aim of the research is to develop an intelligent scheduling system operating locally on the user's equipment without data transmission via Internet. Methods . This paper examines the problem of ensuring the privacy and autonomy of digital assistants dependent on cloud infrastructure. A client-server architecture is proposed, in which the server component is implemented using the FastAPI framework and an SQLite database, and the client interface is written in JavaScript. Schedule visualization and entry editing are performed through the web interface. Results . The system's voice pipeline is described: the Porcupine engine is used for activation, and the Faster-Whisper model with int8 quantization is used for transcription. A comparative analysis of the technology stack, ensuring high speech recognition accuracy, is conducted. A hybrid natural language understanding module is developed. RAG technology is implemented, integrating schedule data into the response generation context. Speech synthesis is performed using the Piper neural network, whose execution through ONNX Runtime ensures high processing speed. A heuristic greedy search algorithm for managing time resources has been developed. Conclusion . The developed system is considered applicable in the corporate sector, where information security and operation in closed network environments are critical.

Precise spatial distribution of interstitial needles is critical for 3D-printing-assisted brachytherapy in cervical cancer. This study proposes a greedy algorithm-based needle trajectory planning (GANTP) framework to generate patient-specific needle configurations while ensuring needle collision avoidance and achieving clinically acceptable high-risk clinical target volume (HR-CTV) coverage in compliance with OAR dose constraints. The GANTP framework comprises three core steps: (1) Generation of candidate trajectories anchored within clinically viable entry zones; (2) Parameter-driven greedy selection of needle trajectories based on a geometric influence radius (δ) evaluated at three discrete values (12, 15, and 18mm), where δ serves as a geometric surrogate for dose coverage, together with a geometric coverage-ratio threshold (γ = 98%) and a collision-free margin (d) relative to the tandem; and (3) Dosimetric evaluation and inverse planning with dwell-time optimization. The framework was evaluated using CT datasets from 20 cervical cancer patients. Performance metrics, including HR-CTV coverage, organs-at-risk (OAR) doses (D2cc), needle counts, and efficiency, were compared against manual planning. GANTP was able to generate clinically acceptable plans for all 20 cases. Based on a final-selection strategy that prioritized clinically acceptable HR-CTV coverage (≥ 90%) and the lowest needle count among the evaluated δ settings, δ = 15mm was selected for 16 patients and δ = 18mm for 4 patients as the final selected configurations. Compared to manual planning (HR-CTV coverage: 92.62 ± 1.51%), these Final Selected plans achieved clinically acceptable coverage of 91.96 ± 1.24% (P = 0.024), consistently exceeding the 90% clinical threshold. The OAR sparing was no significant difference to manual planning: D2cc for the rectum (66.11 ± 3.99Gy vs. 66.46 ± 4.00Gy, P = 0.542), bladder (78.28 ± 5.16Gy vs. 79.45 ± 4.47Gy, P = 0.201), and sigmoid (63.30 ± 4.86Gy vs. 62.33 ± 6.84Gy, P = 0.916). The algorithm significantly reduced the average number of needles from 5.40 ± 0.94 to 4.25 ± 0.55 (P < 0.001). The most substantial improvement was observed in one case (Patient 2), where the needle count was reduced from 7 to 4 while maintaining a coverage of 91.7%. The total planning workflow time was substantially reduced from 2 to 3h (manual) to 8.2 ± 1.4min (GANTP), with algorithm execution taking less than 190s across all δ settings. GANTP establishes a semi-automated, patient-specific framework for generating collision-free, non-coplanar trajectories that meet clinical dosimetric goals with a reduced mean number of needles. Integrated with 3D-printed templates, this approach demonstrates significant potential for improving the precision and efficiency of interstitial brachytherapy. Future work will include phantom experiments for physical validation.

Model trees provide an appealing way to perform interpretable machine learning for both classification and regression problems. In contrast to "classic" decision trees with constant values in their leaves, model trees can use linear combinations of predictor variables in their leaf nodes to form predictions, which can help achieve higher accuracy and smaller trees. Typical algorithms for learning model trees from training data work in a greedy fashion, growing the tree in a top-down manner by recursively splitting the data into smaller and smaller subsets. This yields a fast algorithm, but the selected splits are only locally optimal, potentially rendering the tree overly complex and less accurate than a tree whose structure is globally optimal for the training data. In this paper, we empirically investigate the effect of constructing globally optimal model trees for classification and regression. The trees we consider feature linear support vector machines at the leaf nodes and are learned using mixed-integer linear programming (MILP) formulations. We use benchmark datasets to compare them to model trees obtained using greedy and dynamic programming-based algorithms, evaluating both tree size and predictive accuracy. We also compare to classic optimal and greedily grown decision trees, random forests, and support vector machines. Our results show that MILP-based optimal model trees can achieve competitive accuracy with very small trees. We also investigate the effect on the accuracy of replacing axis-parallel splits with multivariate ones, foregoing interpretability while potentially obtaining greater accuracy.

We present SCITE-RNA, a novel phylogenetic tree inference method designed for single-cell RNA sequencing data which takes reference and alternative read counts of single-nucleotide variants as input. Our approach uses a maximum-likelihood random-scan greedy search that alternates between cell lineage tree and mutation tree representations to escape local optima until convergence is achieved in both. We demonstrate superior performance on simulated data compared to existing methods. Furthermore, we show its applicability to cancer single-cell RNA sequencing data, where it allows us to link evolutionary trajectories of cells to their gene expression profiles.Supplementary InformationThe online version contains supplementary material available at 10.1186/s13059-026-04123-w.

The proliferation of distributed energy resources (DERs) and the ubiquity of Internet of Things (IoT) devices are driving the integration of mobile edge computing (MEC) into smart grids. This convergence enables real-time data processing for prosumers but introduces a complex cyber-physical coupling: computational offloading decisions directly impact local energy consumption, thereby altering the prosumer’s status in the peer-to-peer (P2P) energy market. Conversely, dynamic market prices influence the economic viability of offloading. This paper addresses the joint optimization of computational task offloading and P2P energy trading in an edge-assisted smart grid ecosystem. We formulate the problem as a mixed-integer nonlinear programming (MINLP) model aimed at maximizing long-term system utility, balancing throughput, latency, and economic incentives under strict edge server capacity and community energy neutrality constraints. To tackle the curse of dimensionality and system stochasticity, we propose a hybrid framework combining Deep Q-Networks (DQN) with a constraint-aware heuristic mechanism. The DQN agent learns adaptive offloading policies from high-dimensional states, while a deterministic rule-based layer ensures strict adherence to community energy balance. Simulation results based on real-world solar generation and market data demonstrate that our proposed method outperforms baseline strategies—including local-only execution and greedy heuristics—improving average utility by 12.3% and reducing task delay by 16.5%, while maintaining robust operational feasibility.

Spotify provides audio metadata that can be utilized to support physical activities such as jogging. This study compares the performance of Greedy and Dynamic Programming algorithms for Spotify playlist optimization modeled as a 0/1 Knapsack Problem. Song duration is treated as weight, while a score derived from popularity and energy is used as value. The dataset was obtained from Spotify Wrapped 2025 Top 50 Songs and Spotify All-Time Top 100 Songs, resulting in 31 candidate songs after preprocessing and filtering. Experiments were conducted on playlist durations of 30, 45, 60, 75, and 90 minutes. The results show that Dynamic Programming consistently achieved higher total scores than Greedy across all scenarios. For the 60-minute playlist, Dynamic Programming obtained a total score of 1897 compared to 1894 achieved by Greedy. However, Greedy required a lower execution time (4.244 ms) than Dynamic Programming (16.196 ms). The average optimality gap between the two methods was 1.89%, indicating that Greedy produced solutions that were close to the optimal solutions generated by Dynamic Programming while requiring less computation time.

The rapid evolution of 5G and emerging 6G networks has increased the demand for wireless communication systems that deliver high capacity, low latency, and adaptability. However, conventional terrestrial infrastructure remains costly and inflexible, particularly in dynamic or remote environments. This article develops a new Federated Reinforcement Learning (FRL)-based UAV communication system using Selective Entropy-Fused Proximal Policy Optimization (SEF-PPO) is proposed to enhance the performance of locally-on-policy learning in real-time decision-making environments. In contrast to existing digital twin or offline-trained deep reinforcement learning (DRL) methods, the proposed solution eliminates the need for replay buffers, thereby reducing memory and computational requirements for UAV platforms. UAVs learn collaboratively while preserving data privacy and maintaining robustness to non-IID user distributions through federated aggregation with a High-Altitude Platform (HAP). The framework integrates trajectory planning, user association, energy-efficient resource allocation, and handover management within a unified adaptive architecture. Experimental results demonstrate significant improvements in throughput, fairness, latency, and energy efficiency compared with baseline methods, including DMTD, DRL-EC3, and greedy and random algorithms. Overall, the proposed design enables scalable, energy-aware, and environment-responsive UAV coordination, offering a deployment-ready solution for next-generation wireless networks without requiring simulation-based pretraining.

Rain gauge networks are the core infrastructure for hydrological and water resource monitoring, flood control and disaster mitigation early warning, and water resource planning and regulation. The rationality of their layout directly determines the accuracy, representativeness, and economy of regional precipitation data acquisition. Considering that information entropy can accurately characterize the spatial distribution law and information complexity of rainfall, and spatiotemporal deep learning models have strong capabilities in fitting spatiotemporal features, this paper couples mutual information entropy with a spatiotemporal deep learning model and proposes a novel optimal layout method for rain gauge networks. Daily observed rainfall data from 50 ground-based rain gauges in the upper reaches of the Tuojiang River during 2015–2024, as well as the PERSIANN-CCS remote sensing precipitation product for the same period, were used in the study. A CNN-LSTM spatiotemporal deep learning model integrating spatial features and temporal dependence was constructed, coupled with the mutual information entropy index, and the GA-PSO hybrid optimization algorithm was applied for solution. The superiority of the proposed method was verified by comparison with the calculation results of the traditional mutual information entropy-based greedy optimization algorithm. The results show that the hybrid optimization algorithm driven by the spatiotemporal deep learning model coupled with mutual information entropy is significantly superior to the comparison algorithm in terms of the rationality of the station network structure, the ability to characterize spatial rainfall distribution, the control of average relative error, and the improvement of total information entropy. After optimization, the number of rain gauges in the upper reaches of the Tuojiang River can be reduced from 50 to 25. While greatly reducing the number of stations, the optimized network can still relatively accurately reflect the spatiotemporal characteristics of rainfall in the basin, which can provide a theoretical basis and technical support for the optimal layout of basin rain gauge networks and water resource management.

Anatomic Feasibility of Off-the-Shelf FEVAR Devices: An Anatomic Study.

Since cluster analysis methods usually cannot be applied directly to data with missing values, various approaches have been investigated to handle the problem. Multiple imputation is one of the standard procedures for addressing the problem of missing data. In cluster analysis, instead of Rubin's rule, cluster ensemble methods have been proposed to be combined with multiple imputation. However, it remains unrevealed which of the cluster ensemble algorithms leads to better performance when integrated with the procedure. Therefore, we conducted numerical comparisons of several algorithms to integrate the results from k-means++ clustering for multiply imputed datasets and also applied the combined approaches to two real datasets. Our results suggest that the non-negative matrix factorization algorithm may be suitable for scenarios with class balance, whereas the greedy and agglomerative cluster algorithms may be suitable for scenarios with class imbalance. Before application to actual datasets, we still recommend performing simulation experiments in scenarios reflecting the characteristics of the datasets and the assumption of missing value mechanisms.

A population-based and Q-learning-cooperative iterated greedy algorithm for distributed no-wait job shop problem

Sparse signal recovery for a new hybrid greedy algorithm based mutual coherence

Iterated greedy algorithm with constraint programming for scheduling steelmaking-continuous casting process

Nowadays, closed-circuit television (CCTV) cameras are deployed worldwide to monitor movements of humans and other objects to improve the efficiency and safety of societies. Therefore, their proper placement is crucial for achieving effective surveillance coverage. Additionally, their proper placement is significantly important for maximizing visual coverage while reducing installation/management costs. For this task, digital twin is a useful technology, since it can simulate coverage and blind spots while freely changing camera locations. To implement digital twin, 3D modelling of a structure including a complex room is a key issue. In this paper, we propose a 3D indoor modelling method using 360° panoramic images and show its application to a CCTV camera placement optimization. This method constructs a structured 3D model of a target room from captured 360° panoramic images using a 3D Gaussian Splatting reconstruction method based on a visual simultaneous localization and mapping (VSLAM) framework. The Inertial Measurement Unit (IMU) is used together to improve the camera position estimation accuracy. The model construction is anchored using a GNSS/GPS reference to establish global spatial coordinates. As an application of the generated 3D model, optimal locations of a given number of CCTV cameras are determined by combining ray-casting visibility analysis and a greedy optimization algorithm in the virtual environment, maximizing visual coverage while minimizing blind spots and avoiding excessive overlap between camera views. For evaluations, we applied the proposed method to three rooms in Okayama University, Japan, and seven rooms in the Indonesian Institute of Business and Technology, Indonesia. After optimizing camera locations in the virtual environment, the cameras were actually installed in the rooms according to the recommended positions. The performance was evaluated using visibility coverage, blind spot reduction, and Root Mean Squared Error (RMSE) between the estimated and actual camera positions, where promising results were achieved.

The task scheduling problem in fog computing is one of the key challenges in the development of fog computing within next-generation communication networks. Addressing this challenge requires balancing processing performance with resource constraints while meeting network conditions. Given the distributed and heterogeneous nature, as well as the dynamic topology, optimally allocating tasks to fog nodes is a complex issue. To contribute to solving this problem, we propose a task scheduling method based on an improved Memetic algorithm. The proposed method leverages the strengths of evolutionary algorithms and local search, while incorporating a task restructuring mechanism, to enhance allocation efficiency and task processing in the fog computing environment. Simulation experiments demonstrate that the proposed method outperforms genetic algorithms, Round-Robin, Greedy methods, and the Ant Colony Optimization algorithm in terms of efficiency. This study provides a fresh, simpler approach that aligns with network conditions while still achieving the desired performance. KEYWORDS: 6G, fog computing, task allocation, memetic algorithm, task restructuring mechanism, next-generation networks.

AimsTo examine the relationship between surgical treatment type (acute total hip arthroplasty (THA) vs open reduction and internal fixation (ORIF)) and in-hospital medical complications in older adult trauma patients with operatively managed acetabulum fractures.MethodsWe conducted a retrospective cohort study of patients aged ≥ 50 years who presented to institutions participating in the Trauma Quality Improvement Program between 1 January 2017 and 31 December 2022, and who underwent acetabulum fracture surgery within three weeks of admission. Our primary outcome was the development of in-hospital medical complications. Secondary outcomes included each medical complication alone, hospital length of stay, and discharge disposition. Acute THA patients were matched 1:1 without replacement to patients treated with ORIF on the logit of the propensity score using a greedy nearest-neighbour matching algorithm. Generalized estimating equations were used to calculate percent absolute risk differences with 95% CIs for categorical outcomes in the propensity score-matched sample. Wilcoxon signed-rank tests were used compare within pair differences in continuous outcomes.ResultsA total of 10,213 patients were included in our study, of which 1,226 (12%) were treated with an acute THA and 8,987 (88%) were treated with ORIF. A total of 1,223 acute THA patients were matched to 1,223 ORIF patients. After matching, there were no meaningful differences in any baseline characteristics between the two treatment groups. There was no difference in the risk of in-hospital complications between patients treated with acute THA (216/1,223 (17.7%) vs patients who were treated with ORIF (201/1,223 (16.4%)) (absolute risk difference 1.23%, 95% CI -1.71 to 4.17, p = 0.414). There were no significant differences in the risk of each complication, length of stay, or discharge disposition.ConclusionOur results suggest that acute THA and ORIF demonstrate similar risks of postoperative medical complications among older patients with acetabular fractures.Cite this article: Bone Jt Open 2026;7(5):643–650.

Due to an increasing in online package orders, delivery providers need to determine how to reduce driver overtime while maintaining on-time delivery. To address this, our research introduces the Hybrid Helper Dispatching Problem with Drone (HHDPD), which aims to reduce the truck's return time to the depot and ensure the simultaneous return of both the driver and the drone. This study utilises the OR-Tools package and a greedy algorithm to address the routing problem and selection of drone service areas. Computational experiments are carried out to evaluate the best possible solution among four cases: TSP (Travelling Salesman Problem) without a driver helper, standard helper (including one driver, one independent helper, and one truck), HHDP (Hybrid-Helper Dispatching Problem), and HHDPD. Results demonstrate that the HHDPD model can significantly reduce the overall time spent, minimise idle resource wastage of driver helper and improve package delivery efficiency.

Contemporary large-scale digital platforms face severe challenges in multi-channel resource slot allocation, including cross-channel redundancy, metric fragmentation, fairness-efficiency conflicts, and ignorance of asymmetric cross-space spillover effects, leading to significant losses in revenue and user engagement. This paper proposes AllocOpt, a submodular constrained optimization framework for heterogeneous resource slot allocation in multi-dimensional recommendation landscapes, which explicitly models directed cross-channel spillover effects and integrates efficiency, engagement, and fairness objectives. The framework formalizes cross-channel interactions as a directed acyclic spillover graph, proves the approximate submodularity of the global objective under reasonable constraints, and designs a polynomial-time greedy algorithm with (1−1/e) approximation guarantee. Meanwhile, a contextual Thompson-sampling bandit algorithm is adopted for online learning of spillover parameters, and hierarchical fairness is enforced via stochastic dominance instead of rigid quotas. Validated in 18-month production with over 200 million users, AllocOpt improves allocation efficiency by 38.7%, reduces cross-channel redundancy by 67.4%, enhances user satisfaction by 28.4%, and boosts small merchants’ weekly sales by 128.6%, with only 6.3 ms computational latency per optimization cycle. This framework unifies real-time performance, allocation efficiency and ecosystem fairness, and is generalizable to e-commerce, mobility and other large-scale platforms with heterogeneous recommendation surfaces.

As electric vehicles gain traction across the globe, one persistent worry among drivers is whether their battery will last long enough to reach the next charging point, a concern commonly referred to as range anxiety. In this paper, we describe a practical route planning tool that tackles this problem head-on. At its core sits a Gradient Boosting Regressor trained on 20,000 synthetically generated trip records whose parameters are rooted in real-world physics. The model takes in the vehicle type, how much cargo is on board, the trip distance, driving speed, terrain changes, and outside temperature, and outputs an energy consumption estimate. On the server side, a FastAPI application pulls together driving directions from OSRM, live weather readings from OpenWeatherMap, elevation data from Open-Elevation, and nearby charger locations from OpenChargeMap. A step-by-step greedy algorithm then figures out where the driver should stop to recharge, while also factoring in how much the battery may have degraded over time. The accompanying mobile app, built with Flutter, shows the planned route on an interactive map and even works offline thanks to local caching. In our tests, the prediction model achieved an R² above 0.95 on unseen data.