Search Results (624)

In tennis, where performance is governed by complex kinetic chain interactions, objective skill classification is vital for coaching and talent identification. This study presents a hierarchical deep learning framework leveraging synchronized bilateral Inertial Measurement Unit (IMU) data from 39 participants (11 elite, 28 amateur). The proposed system successfully distinguishes expertise levels across a total of 4594 strokes, including augmented samples.. A hybrid Convolutional Neural Network-Bidirectional Long Short-Term Memory (CNN-BiLSTM) architecture was developed to autonomously extract spatiotemporal features from the raw kinematic signals of forehand, backhand, service, and volley strokes. The proposed model achieved an accuracy of 95.54%, significantly outperforming both traditional machine learning and state-of-the-art deep learning benchmarks. Qualitative t-distributed Stochastic Neighbor Embedding (t-SNE) analyses revealed that elite athletes form highly homogeneous clusters in the feature space. Furthermore, quantitative Asymmetry Index assessments confirmed that professionals exhibit superior bilateral coordination stability. These findings demonstrate that the proposed end-to-end system offers a robust, field-applicable solution for identifying technical excellence. It provides coaches with reliable digital biomarkers, thereby overcoming the limitations of subjective visual observation. Full article

Most artistic expressions are conveyed through images (e.g., painting) and audio (e.g., music), and deep learning has been successfully applied to neural style transfer within each of these modalities. However, there is still a lack of deep models that explicitly learn to transfer style between images and audio. Motivated by synesthesia, which reflects intrinsic connections between vision and hearing in the human brain, we propose a deep hybrid synesthesia model for audio–image style transfer. Our framework consists of two main components: (1) a component conversion module that learns cross-modal mappings between audio rhythm/spectrum and image color/shape in a continuous valence–arousal (VA) emotion space; and (2) a style conversion module that transfers high-level artistic styles between Eastern (ink-wash, shui-mo) and Western painting and their corresponding musical counterparts. We first learn emotion-aware feature networks that align low-level audio and visual components based on shared affective representations, and then model long-term stylistic structures for cross-modal style transfer. Experiments include “seeing the sound” (audio-to-image generation with controllable components) and full audio–image style transformations. Both objective analyses and subjective evaluations suggest that our model can produce cross-modal artworks whose perceived style and emotional content are consistent with human synesthetic impressions. Full article

Deep learning methods have significantly improved hyperspectral image (HSI) classification by exploiting hierarchical feature learning to integrate spatial and spectral information, thus significantly improving classification accuracy. Nevertheless, current deep learning approaches (such as CNNs, Transformers and Mamba) still face three major challenges: inadequate mitigation of spectral redundancy, high computational costs associated with global modeling, and the loss of two-dimensional spatial structure during sequential processing. To address these issues, we propose LMamba, a task-oriented hybrid framework that combines multi-scale convolutional filtering with local-context-conditioned state space modeling for hyperspectral image classification. Rather than introducing a fundamentally new SSM formulation, LMamba focuses on adapting the input-dependent parameter projection of Mamba to HSI data by injecting local 2D neighborhood context into the generation of selective SSM parameters. This design enables the state space module to better preserve spatial continuity while maintaining linear-complexity sequence modeling. The framework consists of two core components. First, the Multi-scale Aggregation and Compression Block (MACB) employs parallel grouped convolutions with varying kernel sizes to capture spatial features at multiple scales while simultaneously reducing spectral redundancy through channel compression. Second, the Locally Guided 2D Scanning Mechanism replaces conventional unidirectional 1D scanning with a context-aware 2D scanning strategy, thereby preserving structural continuity and enhancing feature representation by integrating local neighborhood spatial information into state transitions. Validation on three prominent HSI datasets demonstrates that LMamba consistently outperforms state-of-the-art methods based on CNNs, Transformers, and SSMs as measured by overall accuracy (OA), average accuracy (AA), and the Kappa coefficient. In summary, LMamba provides an efficient and accurate HSI classification framework under the considered benchmark settings, and its compact complexity and low-sample robustness suggest potential usefulness for practical HSI analysis. Full article

Optimal allocation of distributed generation (DG) and feeder reconfiguration are critical strategies for improving the operational efficiency and voltage stability of modern radial distribution networks under increasing penetration of renewable resources. However, the simultaneous optimization of DG placement, sizing, and network topology constitutes a highly nonlinear multi-objective problem subject to electrical, operational, and radiality constraints. Unlike existing studies that treat DG allocation and feeder reconfiguration as separate or weakly coupled problems, this work introduces a unified mixed-integer nonlinear optimization framework that captures their strong interdependency. In addition, a hybrid Big Bang–Big Crunch (HBB-BC) algorithm is proposed, combining stochastic contraction with adaptive learning mechanisms to improve convergence robustness in highly nonlinear search spaces. This contribution addresses the limitations of conventional metaheuristics in handling coupled topology–generation optimization problems and provides a scalable solution for modern active distribution networks. We propose a coordinated optimization framework for optimal DG placement and feeder reconfiguration aimed at minimizing real power losses while enhancing voltage stability and reducing both operational cost and environmental impact. The problem is formulated as a constrained multi-objective optimization model and solved using an improved hybrid Big Bang–Big Crunch metaheuristic algorithm which integrates exploration and exploitation mechanisms to achieve fast convergence and robust global search performance. The proposed method is validated on both IEEE 33-bus and IEEE 69-bus radial distribution systems under multiple operational scenarios. The results demonstrate that the coordinated optimization consistently achieves significant performance improvements across different network scales, confirming the robustness and scalability of the proposed framework. Full article

Maritime emergency response requires broadband and reliable communications in sea areas where shore coverage is limited or emergency connectivity is temporarily unavailable, making rapid on-demand aerial networking essential. Unmanned aerial vehicles (UAVs) acting as aerial base stations can be rapidly deployed to provide on-demand coverage; however, ship mobility, heterogeneous emergency priorities, and UAV endurance limitations make the joint optimization of user association and multi-UAV deployment a challenging mixed-integer, long-horizon decision problem. This paper considers a multi-UAV maritime emergency communication system where ships are categorized into multiple priority classes and served links must satisfy a minimum signal-to-noise ratio (SNR) constraint. We formulate a long-term system-utility maximization problem that jointly determines (i) per-slot association between UAVs and ships under capacity, priority, and SNR constraints, and (ii) dynamic UAV deployment under mobility, geofencing, and battery constraints. To obtain tractable and high-quality solutions, we decompose the problem into two coupled subproblems. For user association, we propose a Priority-Aware Branch-and-Cut (PA-BAC) algorithm that integrates linear programming relaxation, cutting-plane tightening, and priority-guided branching, with a priority-greedy feasible initialization to accelerate incumbent improvement. For dynamic deployment, we develop an Enhanced Multi-Agent Proximal Policy Optimization (E-MAPPO) method featuring a global value network, entropy regularization, and sequential actor updates to enhance learning stability and exploration. Importantly, the PA-BAC association is embedded into the learning loop to provide reliable, constraint-satisfying per-slot rewards and reduce the burden of end-to-end learning over hybrid-action spaces. Simulation results demonstrate that PA-BAC consistently improves normalized priority-weighted throughput over heuristic association baselines. Moreover, by mathematically enforcing priority and QoS feasibility at every slot and delegating only continuous mobility to MARL, the integrated E-MAPPO-PA-BAC framework achieves higher long-term system utility, improved energy efficiency, and strong robustness across varying ship densities—properties that are vital for time-sensitive maritime emergency communications. Additional runtime, sensitivity, and AIS-driven trace evaluations further verify the computational practicality of PA-BAC and the applicability of the proposed framework under realistic ship mobility patterns. Full article

Predicting financial stock returns remains challenging due to their inherent nonlinearity, non-stationarity, and sensitivity to market microstructure noise. Existing approaches typically rely on either econometric filtering techniques or deep learning models in isolation, limiting their ability to jointly capture latent dynamics and complex temporal dependencies. This study proposes a hybrid Extended Kalman Filter–Long Short-Term Memory (EKF–LSTM) framework that integrates nonlinear state-space filtering with deep sequential learning. The EKF component performs nonlinear state estimation and denoises to extract latent signals from noisy observations, while the LSTM network models nonlinear temporal dependencies in the filtered series. The proposed framework is evaluated using data from multiple international markets, including China, the United States, and Europe, providing cross-market evidence of model robustness. Empirical results show that the EKF–LSTM model consistently outperforms benchmark models (ARIMA, standalone EKF, LSTM, and GRU) across standard statistical metrics, including RMSE, MAE, and mean directional accuracy (MDA). In addition, the model delivers economically meaningful improvements under a long-only trading strategy, achieving higher risk-adjusted returns and lower maximum drawdowns relative to benchmark strategies. Diebold–Mariano tests further confirm that these performance gains are statistically significant. Overall, the findings demonstrate that integrating nonlinear state-space filtering with deep learning provides a robust and effective framework for financial time-series forecasting. However, the results should be interpreted with caution due to the limited sample size and the simplifying assumptions underlying the trading strategy. Full article

Forecasting scrap steel prices is crucial for the economic sustainability of recycling operations, yet it remains challenging due to inherent volatility and non-stationary behavior. In this study, we develop and evaluate a delta-targeted Hybrid forecasting pipeline for short horizons of 1, 3, and 7 days. We benchmark classical baselines (Naive, Seasonal Autoregressive Integrated Moving Average (SARIMA), and Exponential Smoothing (ETS)) against recurrent deep learning models (Simple Recurrent Neural Network (RNN), Gated Recurrent Unit (GRU), and Long Short-Term Memory (LSTM)) and recent neural forecasting baselines, including Decomposition-Linear (DLinear), Convolutional Kolmogorov–Arnold Network (C-KAN), and Neural Basis Expansion Analysis for Time Series (N-BEATS), using real-world daily scrap steel price data. The results indicate that delta-targeting generally yields more stable predictive performance than direct raw-price forecasting as the prediction horizon increases. For example, at the 7-day horizon, the predictive fit improves from approximately $R^2\approx 0.87$ for raw-price LSTM to around $R^2\approx 0.90$ for delta-trained recurrent models. At the same horizon, a delta-based RNN achieves the lowest Mean Absolute Percentage Error (MAPE) among the evaluated models (approximately $1.39\%$), while the proposed Hybrid model remains competitive across all tested horizons and maintains a goodness-of-fit of approximately $R^2\approx 0.90$ without uniformly minimizing point error relative to the best-performing recurrent baseline. Attention profiling and permutation-based feature importance analyses indicate that the model places relatively higher weight on calendar-related inputs, consistent with the presence of weekly patterns in the data; these results should be interpreted as sensitivity diagnostics rather than causal evidence. Overall, the findings suggest that delta-transformed targets provide a more suitable prediction space than raw-price targets for short-horizon scrap steel forecasting, while the Hybrid design offers a balanced combination of predictive performance and diagnostic interpretability for operational decision support. Full article

With the increasing scale and complexity of network traffic, anomaly detection faces significant challenges, particularly under the scarcity of labeled data in real-world environments. Although graph neural networks (GNNs) effectively model relational structures, most existing approaches rely on supervised learning, limiting their applicability in weakly labeled or unlabeled scenarios. To address these limitations, this paper proposes a self-supervised graph neural network framework, termed EGSCA, for network traffic anomaly detection. The framework employs a GNN to jointly model node and edge information, enabling the learning of discriminative representations. On this basis, a graph contrastive learning strategy is designed, where diverse subgraphs are generated via breadth-first search (BFS) to effectively capture local structural patterns. Meanwhile, a hybrid contrastive loss based on Wasserstein distance and Gromov–Wasserstein distance is introduced to achieve collaborative optimization between feature-space alignment and structural consistency under unlabeled conditions. Experimental results on multiple benchmark datasets demonstrate that the proposed method achieves competitive performance. Notably, it achieves the best results on datasets NF-BoT-IoT and NF-BoT-IoT-v2, with average improvements of approximately 3.2% in F1-score and 1.7% in DR over the strongest baseline. Further analysis indicates that the model yields more pronounced performance gains in scenarios with high class separability. Full article

This paper proposes a novel hybrid deep learning framework for direction-of-arrival (DOA) estimation using a uniform linear array. Direction of Arrival estimation is a fundamental problem in array signal processing with critical applications in radar, sonar, wireless communications, and speech processing. Traditional methods like MUSIC and ESPRIT provide high resolution but suffer from high computational complexity and poor performance in low signal-to-noise ratio (SNR) environments. Recent advances in deep learning have shown promise in improving DOA estimation accuracy and robustness. The framework synergistically combines a ResNet-based feature extractor with a Mamba state-space model through a feature fusion mechanism. The ResNet branch extracts high-level spatial features from the covariance matrix, while the Mamba branch captures long-range dependencies and sequential patterns. These complementary features are fused and then passed to an MLP for DOA regression. Extensive experiments on simulated datasets demonstrate that, at low SNRs, our fusion model significantly outperforms traditional methods such as MUSIC and ESPRIT, as well as other baseline models, in terms of both estimation accuracy and computational efficiency. Quantitatively, at SNR = −5 dB, the proposed method reduces the RMSE by 41.6% compared to MUSIC. Full article

Historically, pharmaceutical formulation development relied heavily on trial-and-error experimentation, which was useful for empirical progress but often provided limited mechanistic understanding and insufficient efficiency for increasingly complex drug products. The introduction of Design of Experiments (DoE) and Quality by Design (QbD) established a more systematic framework for studying formulation variables, manufacturing parameters, and Critical Quality Attributes (CQAs). Approaches such as factorial designs, response-surface methodology, and mixture designs have therefore become central to modern pharmaceutical development because they improve experimental efficiency and support the definition of design space. However, as formulations become more nonlinear, high-dimensional, and multi-objective, these classical approaches may no longer be sufficient on their own. This review examines the evolution of experimental design in pharmaceutical research, from one-factor-at-a-time experimentation to structured DoE/QbD strategies, and then to emerging intelligent optimization methods. Its central objective is to clarify when conventional DoE/QbD remains appropriate and when it should be complemented by machine learning, Bayesian optimization, digital twins, and closed-loop experimental systems. The review first summarizes the foundations and strengths of classical experimental design; then, it discusses its practical limitations in complex formulation settings, and finally evaluates how data-driven and hybrid approaches can extend pharmaceutical development. Evidence from tablets, capsules, nanocarriers, transdermal patches, and biotherapeutic systems suggests that intelligent optimization can improve predictive performance and experimental efficiency when used alongside, rather than instead of, established pharmaceutical development principles. Full article

Accurate multi-day sea surface temperature (SST) prediction at sub-mesoscale resolution is challenging due to nonlinear ocean dynamics, heterogeneous multi-source observations, and error accumulation during autoregressive rollout. This paper proposes a hybrid Mamba–ConvLSTM framework that combines recurrent local spatiotemporal encoding with selective state-space long-range spatial modeling. The ConvLSTM branch captures local spatial patterns and short-range temporal dependencies through convolutional gating, while the Mamba branch captures long-range spatial dependencies across each frame through cross-direction window scanning and maintains temporal coherence via persistent hidden states across successive time steps. A physically informed preprocessing stage aligns 0.083° reanalysis variables to the 0.05° OSTIA target grid via a Grow-and-Cut strategy and extracts gradient-based advection and diffusion proxy features under boundary-aware finite differencing. During autoregressive rollout, auxiliary variables are held at their last observed values and physical proxies are recomputed from the predicted SST, following a clearly specified protocol. Experiments on a South China Sea benchmark compare the proposed model against nine baselines—including persistence, daily climatology, ConvLSTM, PredRNN, ConvGRU, TCTN, PANN, Swin-UNet, and ViT-ST—under an identical data-split, normalization, and rollout protocol. Evaluation with RMSE, MAE, SSIM, $R^2$, and anomaly correlation coefficient (ACC) shows that the proposed model achieves a 10-day average RMSE of 0.512 °C, outperforming the strongest learning-based baseline ViT-ST by 5.0% and the persistence forecast by 21.0%. Ablation studies, sensitivity analyses, seasonal evaluation, and statistical significance testing verify the contribution of each component and the robustness of the results. Full article

The rapid proliferation of Internet of Things (IoT) devices has significantly expanded the attack surface of modern networks leading to a surge in IoT-based botnet attacks. Detecting such attacks remains challenging due to the high dimensionality and heterogeneity of IoT network traffic. This study proposes and evaluates three hybrid deep learning architectures for IoT botnet detection that combine representation learning with supervised classification: VAE-encoder-MLP, VAE-encoder-GAT, and VAE-encoder-MoTE. A Variational Autoencoder is initially trained to learn a compact latent representation of the high-dimensional traffic features. Subsequently, the pretrained VAE-encoder component is employed to project the data into a lower-dimensional embedding space. These embeddings are then used to train three different downstream classifiers: a multilayer perceptron (MLP), a graph attention network (GAT), and a mixture of tiny experts (MoTE) model. To further enhance representation discriminability, supervised contrastive learning is incorporated to encourage intra-class compactness and inter-class separability. The proposed architectures are evaluated on two widely studied benchmark datasets—the CICIoT2022 and N-BaIoT dataset—under both binary and multiclass classification settings. Experimental results demonstrate that all three models achieve near-perfect performance in binary attack detection, with accuracy exceeding 99.8%. In the more challenging multiclass scenario, the VAE-encoder-MLP model achieves the best overall performance, reaching accuracies of 98.55% on CICIoT2022 and 99.75% on N-BaIoT. These findings provide insights into the design of efficient and scalable deep learning architectures for IoT intrusion detection. Full article

As the number of unmanned aerial vehicles (UAVs) and the volume of computational tasks increase in UAV ad hoc networks (UAVANET), the solution space for task allocation strategies grows exponentially. In practical emergency scenarios with concurrent multi-user access, multi-UAV systems equipped with mobile edge computing (MEC) devices face challenges such as limited computing resources and imbalanced task distribution during task offloading. To address these challenges, this paper proposes an adaptive task allocation algorithm named AUSTA-DQHO (Adaptive UAV Swarm Task Allocation using Deep Q-networks and Genetic Algorithms Hybrid Optimization), which combines Deep Q-Network (DQN) with Genetic Algorithm (GA), aiming to optimize computational task scheduling and minimize both the total task delay and the variance in task delays. First, we introduce a multi-UAV-assisted MEC application framework. In this framework, UAVs equipped with high-performance computing modules are deployed as airborne servers in the target area, providing data offloading and task computation support for IoT devices. Next, to tackle the optimization problem, we replace the random action selection process in DQN with a hybrid strategy that incorporates heuristic methods—specifically, GA and greedy algorithms—to perform global search and make more effective decisions for optimal task allocation for each offloading request. Furthermore, to accelerate the convergence of the AUSTA-DQHO policy while ensuring global optimality, we introduce a pre-clustering mechanism and a dynamic weighting factor for randomly generated task offloading requests in the target area. These mechanisms effectively reduce the solution space and ensure that optimal actions are learned at different stages of the training process. Experimental results demonstrate that the proposed algorithm achieves a task latency reduction of 18.72% and a load balancing improvement of 98.72%, surpassing the performance of the other algorithms. Additionally, we explore the optimal number of UAVs under given environmental conditions to minimize the waste of computing resources. Full article

Reducing cement consumption in mortar systems is essential for lowering the environmental impact of cement-based materials. Conventional mix design approaches rely mainly on particle size distribution and fineness modulus, which do not fully capture the effects of aggregate packing, morphology, and petrographic composition on paste demand and mechanical performance. Fourteen fine aggregates of distinct geological origins were experimentally characterized in terms of physical and petrographic properties. A dataset of 211 mortar mixtures, yielding 633 transverse-strength observations, was used to train a Random Forest Regressor (RFR) model for strength prediction. The model achieved $R^2=0.762$ (RMSE = 0.223 kN; MAE = 0.165 kN), demonstrating its reliability as a surrogate screening tool. This study presents a hybrid framework that integrates particle packing theory with machine learning to optimize fine aggregate blends. By introducing a Paste Demand Index (PDI)—combining normalized uncompacted void content, surface texture, and shape—the framework enables the identification of mixtures that minimize paste demand while maintaining mechanical performance under strength constraints. Results confirm that the proposed PDI and strength-based filtering are robust, offering a physically grounded decision-support methodology for narrowing the design space. Ultimately, this approach provides an efficient strategy for resource optimization, effectively bridging the gap between computational screening and laboratory validation in cement-reduction initiatives driven by the cement-based tile manufacturing industry. Full article

Accurate prediction of wake interactions in twin vertical-axis hydroturbine (VAHT) arrays is important for dense tidal-farm layout assessment but remains computationally expensive when based directly on Computational Fluid Dynamics (CFD) reference simulations. While simplified analytical models offer speed, they fail to capture the non-axisymmetric wake characteristics of VAHT arrays, and standard Physics-Informed Neural Networks (PINNs) often struggle with convergence in small-sample, high-dimensional flow settings. To address this challenge, this study proposes a Physics-Informed POD-PINN framework for predicting configuration-wise time-averaged wake fields. The hybrid architecture combines Proper Orthogonal Decomposition (POD) for dimensionality reduction with a dual-branch neural network: a global POD branch captures dominant flow structures, while a lightweight spatial correction branch acts as a continuity-informed regularization on the predicted field. Trained on CFD-generated reference data covering diverse longitudinal and lateral spacing configurations, the model learns to map geometric parameters to a three-component wake field represented on a regularized 3D grid. Results show that the proposed framework achieves the lowest mean streamwise error among the tested surrogate models while maintaining millisecond-level inference speed. This study provides an efficient and physics-aware surrogate tool for repeated wake-field evaluation in twin-hydroturbine configuration exploration. Full article