Search Results (7,859)

Pruning and acceleration has become an essential and promising technique for convolutional neural networks (CNN) in remote sensing image processing, especially for deployment on resource-constrained devices. However, how to maintain model accuracy and achieve satisfactory acceleration simultaneously remains to be a challenging and valuable problem. To break this limitation, we introduce a novel pruning pattern of filter-wise mask by enforcing extra filter-wise structural constraints on pattern-based pruning, which achieves the benefits of both unstructured and structured pruning. The newly introduced filter-wise mask enhances fine-grained sparsity with more hardware-friendly regularity. We further design an acceleration architecture with optimization of calculation parallelism and memory access, aiming to fully translate weight pruning to hardware performance gain. The proposed pruning method is firstly proven on classification networks. The pruning rate can achieve 75.1% for VGG-16 and 84.6% for ResNet-50 without accuracy compromise. Further to this, we enforce our method on the widely used object detection model, the you only look once (YOLO) CNN. On the aerial image dataset, the pruned YOLOv5s achieves a pruning rate of 53.43% with a slight accuracy degradation of 0.6%. Meanwhile, we implement the acceleration architecture on a field-programmable gate array (FPGA) to evaluate its practical execution performance. The throughput reaches up to 809.46MOPS. The pruned network achieves a speedup of 2.23× and 4.4×, with a compression rate of 2.25× and 4.5×, respectively, converting the model compression to execution speedup effectively. The proposed pruning and acceleration approach provides crucial technology to facilitate the application of remote sensing with CNN, especially in scenarios such as on-board real-time processing, emergency response, and low-cost monitoring. Full article

Aims: To ask UK women with diabetes whether they have discussed sexual health with healthcare professionals (HCPs) during diabetes care, and to explore communication barriers. Methods: An online questionnaire was developed, based on a published HCP communication survey, piloted by six women with diabetes. A total of 163 participants, recruited via social media and HCP network, completed Part 1 by selecting Likert or narrative response options, providing descriptive data. We report proportions with 95% confidence intervals (Wilson); percentages are calculated using the number responding to each item. Item-level missingness is retained as a non-analysed category, and the n is reported per question. No inferential comparisons were planned a priori. After Part 1 completion, participants could choose to finish, or to continue to Part 2 questions regarding vulval anatomy, function, and vocabulary (77 completed 2A: 80 completed 2B). Part 2 data was analysed thematically. Results: During diabetes care, a minority of participants, 44/163 (27%), said they had ever discussed sexual health, or had been advised how to access sexual health support, 28/163 (17%). If an HCP discussed sexual health, many women said they expected to feel surprised, 114/163 (70%), or pleased, 88/163 (54%). Some participants said they expected HCPs would find the topic inappropriate, 56/163 (36%), or annoying, 44/163 (27%). Some participants expressed HCP gender preference (75/163 [46%] female and 4/163 [3%] male) for such discussion. Part 2 findings revealed unmet sexual health literacy needs with potential to impact on communication with HCPs. Conclusions: Women reported infrequent communication about sexual health and diabetes during diabetes care. Findings highlight potential communication barriers for some participants including the following: unmet educational needs regarding diabetes and sexual health, lack of confidence about available support, fear of a negative HCP response, and preference for the gender of the HCP. Whereas in previous research, HCPs feared upsetting women by discussing sexual health, many participants said they expected to respond positively. Full article

The swift advancement of renewable energy technology has highlighted the need for effective photovoltaic (PV) solar energy tracking systems. Deep learning (DL) has surfaced as a promising method to improve the precision and efficacy of photovoltaic (PV) solar tracking by utilising complicated patterns in meteorological and PV system data. This systematic literature review (SLR) seeks to offer a thorough examination of the progress in deep learning architectures for photovoltaic solar energy tracking over the last decade (2016–2025). The review was structured around four research questions (RQs) aimed at identifying prevalent deep learning architectures, datasets, performance metrics, and issues within the context of deep learning-based PV solar tracking systems. The present research utilised SLR methodology to analyse 64 high-quality publications from reputed academic databases like IEEE Xplore, Science Direct, Springer, and MDPI. The results indicated that deep learning architectures, including Convolutional Neural Networks (CNNs), Long Short-Term Memory (LSTM) networks, and Transformer-based models, are extensively employed to improve the accuracy and efficiency of photovoltaic solar tracking systems. Widely utilised datasets comprised meteorological data, photovoltaic system data, time series data, temperature data, and image data. Performance metrics, including Mean Absolute Error (MAE), Mean Squared Error (MSE), and Mean Absolute Percentage Error (MAPE), were employed to assess model efficacy. Identified significant challenges encompass inadequate data quality, restricted availability, high computing complexity, and issues in model generalisation. Future research should concentrate on enhancing data quality and accessibility, creating generalised models, minimising computational complexity, and integrating deep learning with real-time photovoltaic systems. Resolving these challenges would facilitate advancements in efficient, reliable, and sustainable photovoltaic solar tracking systems, hence promoting the wider adoption of renewable energy technology. This review emphasises the capability of deep learning to transform photovoltaic solar tracking and stresses the necessity for interdisciplinary collaboration to address current limitations. Full article

This research addresses the complex challenge of integrating modern public transport into historic medieval city centers. These unique urban environments are characterized by narrow streets, protected heritage status, and topographical constraints, which are incompatible with conventional transit vehicles. The introduction of standard bus routes often aggravates traffic congestion and fails to meet the specific mobility needs of residents and visitors. This paper suggests that autonomous electric buses represent a viable and sustainable solution, capable of navigating these constrained environments while aligning with modern energy efficiency goals. The central challenge lies in the optimal selection of an autonomous electric bus that can operate safely and efficiently within the tight streets of historic city centers while satisfying the travel demands of passengers. To address this, a comprehensive study was conducted, analyzing resident mobility patterns—including key routes and hourly passenger loads—and the specific geometric constraints of the road network. Based on this empirical data, a vehicle dynamics model was developed in Matlab^®. This model simulates various operational scenarios by calculating the instantaneous forces (rolling resistance, aerodynamic drag, inertial forces) and the corresponding power required for different electric bus configurations to follow pre-established speed profiles. The core of this research is an optimization analysis, designed to identify the balance between minimizing total energy consumption and maximizing the quality of passenger service. The findings provide a quantitative framework and clear procedures for urban planners to select the most suitable autonomous transit system, ensuring that the chosen solution enhances mobility and accessibility without compromising the unique character of historic cities. Full article

Efficient spectrum allocation and telecom traffic steering in densified heterogeneous maritime communication networks remains a critical challenge due to user mobility, dynamic interference, and congestion across terrestrial, aerial, and sea-based transmitters. This paper introduces the Moving Colorable Graph (MCG) framework, a dynamic graph-theoretical representation of interferences that extends conventional graph coloring to capture the spatiotemporal evolution of heterogeneous wireless links under varying channel and traffic conditions. The formulated spectrum allocation problem is inherently non-convex, as it involves discrete frequency assignments, mobility-induced dependencies, and interference coupling among multiple transmitters and users, thus requiring suboptimal yet computationally efficient solvers. The proposed approach models resource assignment as a time-dependent coloring problem, targeting to optimally support users’ diverse demands in dense port-area networks. Considering a realistic port-area network with coastal, sea and Unmanned Aerial Vehicle (UAV) radio coverage, we design and evaluate three MCG-based algorithm variants that dynamically update frequency assignments, highlighting their adaptability to fluctuating demands and heterogeneous coverage domains. Simulation results demonstrate that the selective reuse-enabled MCG scheme significantly decreases network outage and improves user demand satisfaction, compared with static, greedy and heuristic baselines. Overall, the MCG framework may act as a flexible scheme for mobility-aware and congestion-resilient resource management in densified and heterogeneous maritime networks. Full article

This paper examines how artificial intelligence (AI) can act as a co-creation partner to foster social equity leading to social sustainability by addressing persistent barriers faced by women and minority entrepreneurs. We develop a theoretical framework integrating social capital theory and the resource-based view to analyze how AI can systematically address resource gaps across structural, relational, and cognitive dimensions while serving as a strategic capability that enables competitive advantage. Modern AI systems including ChatGPT, Claude, Gemini, and Perplexity represent practical technologies already operational for everyday entrepreneurs through accessible platforms, low-cost subscriptions, and no-code tools enabling workflow automation with minimal technical skill. While prior work has explored how social capital creates competitive advantages, little research explains how AI technologies specifically enhance both social capital development and resource-based competitive advantage simultaneously for ventures of underrepresented entrepreneurs. This study explicitly identifies the entrepreneurial venture as the unit of analysis and articulates five testable propositions on AI’s influence across structural, relational, and cognitive capital, clarifying mechanisms by which AI functions as a technological mediator that democratizes access to both network resources and strategic capabilities for underrepresented founders. Using AI-generated hypotheticals from Los Angeles demonstrating replicable processes with current technologies like retrieval-augmented generation and cloud AI workspaces, we show that AI-enhanced social capital can reduce venture development disparities while generating distinctive advantages for strategically adopting entrepreneurs. The framework requires empirical validation through longitudinal studies and acknowledges dependencies on infrastructure, ecosystem support, and cultural context, ultimately reconceptualizing AI as an active partner, illustrating that equity and competitive excellence are complementary and achievable through deliberate AI-enabled social capital development. Full article

Existing methods plan the distribution network and sub-transmission network separately. This paper proposes a collaborative renewable energy resource siting and sizing planning method for distribution and sub-transmission networks to increase the renewable energy ratio in high-load density industrial parks and promote the hosting capacity of the power grid. First, to accurately measure planning effectiveness, a renewable energy ratio calculation method is proposed, which comprehensively considers the contributions of green electricity from the power grid and renewable energy generation inside and outside the industrial park. Second, a collaborative planning model is proposed, which optimizes access points and access capacity in the distribution and sub-transmission networks for renewable energy around the park. The net load is better matched with the output of renewable energy outside the park through demand response, thereby maximizing the utilization of the park load to host more renewable energy. Finally, the proposed method is verified in a real industrial park. The method outperforms traditional planning methods in terms of renewable energy ratio in the park and renewable energy hosting capacity outside the park. Full article

Cell-free (CF) networks, with their distributed architecture of access points, offer considerable potential for improving spectral efficiency and expanding coverage. However, the need for dense access point deployment leads to high infrastructure cost and energy consumption. This paper incorporates active reconfigurable intelligent surfaces (RISs)—a low-cost and energy-efficient technology—into cell-free multiple-input multiple-output (MIMO) systems to tackle these challenges and enhance network capacity. Unlike existing active RIS schemes, the proposed method optimizes the spatial configuration of the active elements under a fixed panel layout, harnessing element-level spatial freedom to suppress interference and improve system capacity. We establish a joint optimization framework for active element selection and precoding aimed at maximizing the weighted sum-rate (WSR). An adaptive tabu search (ATS) algorithm is applied to optimize the element topology, and a Lagrangian dual reformulation (LDR) method is introduced to handle the precoding optimization. Simulation results indicate that at a transmit power of $0\mathrm{dBm}$, the passive RIS yields only a $62.49\%$ gain over the no-RIS baseline due to multiplicative fading, whereas the conventional active RIS achieves a $217.46\%$ improvement and the proposed optimized deployment-active RIS further increases the gain to $269.43\%$; thus, our scheme delivers the most significant performance enhancement. Full article

Federated recommendation (FedRec) aims to provide effective recommendation services while preserving user privacy. However, in a federated setting, a single user cannot access other users’ interaction data. With limited local interactions, existing FedRec models struggle to fully exploit interaction information to learn users’ preferences. Moreover, training recommendation models in decentralized FedRec scenarios suffer from a risk of overfitting. To address the above issues, we propose a federated recommendation system with a dual-layer multi-head attention network and regularization strategy (FedDMR). First, FedDMR initializes clients’ local recommendation models. Subsequently, clients perform local training based on their private data. Our dual-layer multi-head attention network is designed to perform attention-weighted interactions on user and item embeddings, progressively capturing local interaction information and generating interaction-aware embeddings, thereby enriching users’ feature representations for modeling personalized preferences. Then, a regularization strategy is employed to guide updates to clients’ models by constraining their deviation from the global parameters, which effectively mitigates overfitting caused by limited local data and enhances the generalizability of the models. Finally, the server aggregates the clients’ uploaded parameters for this round. The entire training process is implemented through the federated learning framework. Experimental results on three datasets demonstrate that FedDMR achieves an average improvement of 2.63% in AUC and precision compared to the recent federated recommendation baselines. Full article

Massive machine-type communications (mMTC) in future 6G networks will involve a vast number of devices with sporadic traffic. Grant-free access has emerged as an effective strategy to reduce the access latency and processing overhead by allowing devices to transmit without prior permission, making accurate active user detection and channel estimation (AUDCE) crucial. In this paper, we investigate the joint AUDCE problem in wideband massive access systems. We develop an innovative channel prior model that captures the dual correlation structure of the channel using three state variables: active indication, channel supports, and channel values. By integrating Markov chains with coupled Gaussian distributions, the model effectively describes both the structural and numerical dependencies within the channel. We propose the correlated hybrid message passing (CHMP) algorithm based on Bethe free energy (BFE) minimization, which adaptively updates model parameters without requiring prior knowledge of user sparsity or channel priors. Simulation results show that the CHMP algorithm accurately detects active users and achieves precise channel estimation. Full article

In this study, we employed functional near-infrared spectroscopy (fNIRS) to explore dynamic functional connectivity (dFC) responses to olfactory stimulation in thirteen healthy control participants and seven patients with vascular dementia (VD). Participants underwent five rest and odor exposure cycles, and dFC was estimated using a sliding window correlation approach. The healthy control group exhibited limited changes, while the VD group exhibited more extensive fluctuations in both oxy- and deoxyhemoglobin dFC across multiple regions during several stimulation periods. Between-group analyses revealed differences, particularly during olfactory stimulation, with moderate to large effect sizes. These preliminary findings suggest that olfactory-evoked dFC may reflect altered brain network dynamics in VD and could potentially serve as a non-invasive, accessible tool to help understand vascular dementia. Full article

Integrating IoT devices into smart grids raises some hard problems related to safe data sharing, the ability to grow, and energy use. Blockchain provides a safe way to check identities without a central authority. Typical ways to confirm transactions, like Proof-of-Work (PoW), use a lot of power, making them bad for devices that cannot use much energy. This study introduces a safe authentication system using Blockchain, a Deep Neural Network (DNN), and a power-saving way to confirm transactions (EACM). The system picks validators based on how much power they have left and their trust scores to save power during confirmation. Using the IoT-Enabled Smart Grid Dataset, simulations showed a transaction speed of 372 TPS, which is 32% better than normal methods. The system correctly authenticates 98.69% of the time, with a confirmation delay of 5.9 milliseconds and an 18% drop in validator node energy use. Also, the system spots 98.4% of unauthorized access tries, with a false acceptance rate (FAR) of 1.7% and a false rejection rate (FRR) of 0.31%. These outcomes prove the system’s ability to offer safe, fast, and energy-saving authentication for big, real-time Smart Grid IoT setups. Full article

Human microglia are central regulators and actors in brain infections and neuro-inflammatory pathologies. However, access to such cells is limited, and studies systematically mapping the spectrum of their inflammatory states are scarce. Here, we generated microglia-like cells (MGLCs) from human induced pluripotent stem cells and characterized them as a robust, accessible model system for studying inflammatory activation. We validated lineage identity through transcriptome profiling, revealing selective upregulation of microglial signature genes and enrichment of microglia/macrophage-related gene sets. MGLCs displayed distinct morphologies and produced stimulus- and time-dependent cytokine secretion profiles upon exposure to diverse inflammatory stimuli, including pro-inflammatory cytokines (TNFα, interferon-γ) and agonists of the Toll-like receptors TLR2 (FSL-1), TLR3 (Poly(I:C)), TLR4 (lipopolysaccharide, LPS), and TLR7 (imiquimod). Transcriptome profiling and bioinformatics analysis revealed distinct activation signatures. Functional assays demonstrated stimulus-specific engagement of NFκB and JAK-STAT signaling pathways. The shared NFκB nuclear translocation response of TLR ligands and TNFα was reflected in overlapping transcriptome profiles: they shared modules (e.g., oxidative stress response and TNFα-related signaling) identified by weighted gene co-expression network analysis. Finally, the potential consequences of microglia activation for neighboring cells were studied on the example of microglia-astrocyte crosstalk. The capacity of MGLC supernatants to stimulate astrocytes was measured by quantifying astrocytic NFκB translocation. MGLCs stimulated with FSL-1, LPS, or Poly(I:C) indirectly activated astrocytes via a strictly TNFα-dependent mechanism, highlighting the role of soluble mediators in the signal propagation. Altogether, this platform enables a dissection of microglia activation states and multi-parametric characterization of subsequent neuroinflammation. Full article

Diabetes is a significant global health issue that demands accurate, accessible, and non-invasive diagnostic methods for effective prevention and treatment. Conventional diagnostic systems are often expensive, painful, time-consuming, and not universally available. In this study, we present a smart system for acetone estimation using simulated breath and a recurrent neural network (RNN) model. The detection system employs a new sensor fabricated from a composite of 1D nanostructured KWO (K₂W₇O₂₂) and 2D nanosheet MXene (Ti₃C₂), designed to measure the chemiresistive response to acetone by mimicking human breath. Resistance data collected by the sensor are used to compute sensitivity values for each acetone concentration (in parts per million, PPM). These values serve as input features for the RNN model, which learns to evaluate health as healthy, high-risk, or diabetic. Trained on acetone concentrations ranging from 0.4 to 2 PPM, the RNN achieves an R² of 99.41% in predicting potential for accurate breath acetone prediction. In future work, we aim to develop a smart device and mobile application based on this model to facilitate real-time diabetes monitoring and prediction. Full article

In this study, we developed an energy-efficient multi-user-associated optimization method involving a massive multi-input multi-output (M-MIMO) system-enabled Cloud Radio Access Network (C-RAN) in Full-Duplex (FD) mode. Maximization of energy efficiency (EE) was achieved with user association. We compose the non-convex multi-objective optimization (MOO) problem for resource allocation and user association in C-RAN. The resultant non-convex MOO problem is non-deterministic polynomial (NP) hard. To tackle this complexity, we find a trade-off between achievable rate and energy consumption. We first reaffirm the problem as an MOO targeting high throughput and minimizing energy consumption instantaneously. By using the epsilon (ε)-constraint method, we transform MOO to an equivalent single objective optimization (SOO) problem by majorization–minimization (MM) approach that enables the transformation of binaries into continuous variables. Further, we propose a multi-objective resource allocation algorithm to obtain a Pareto optimal solution. The simulation results show a significant gain in EE of C-RAN achieved through our proposed MOO algorithm. Our results also show remarkable trade-offs between EE and spectral efficiency (SE). Full article