Search Results (1,587)

Financial time-series prediction remains a significant challenge, driven by market volatility, nonlinear dynamic characteristics, and the complex interplay between quantitative indicators and investor sentiment. Traditional time-series models (e.g., ARIMA and GARCH) struggle to capture the nuanced sentiment in textual data, while static deep learning integration methods fail to adapt to market regime transitions (bull markets, bear markets, and consolidation). This study proposes a hybrid framework that integrates investor forum sentiment analysis with adaptive deep reinforcement learning (DRL) for dynamic model integration. By constructing a domain-specific financial sentiment dictionary (containing 16,673 entries) based on the sentiment analysis approach and word-embedding technique, we achieved up to 97.35% accuracy in forum title classification tasks. Historical price data and investor forum sentiment information were then fed into a Support Vector Regressor (SVR) and three Transformer variants (single-layer, multi-layer, and bidirectional variants) for predictions, with a Deep Q-Network (DQN) agent dynamically fusing the prediction results. Comprehensive experiments were conducted on diverse financial datasets, including China Unicom, the CSI 100 index, corn, and Amazon (AMZN). The experimental results demonstrate that our proposed approach, combining textual sentiment with adaptive DRL integration, significantly enhances prediction robustness in volatile markets, achieving the lowest RMSEs across diverse assets. It overcomes the limitations of static methods and multi-market generalization, outperforming both benchmark and state-of-the-art models. Full article

This exploratory study investigates preliminary trends in the optimization of mechanical properties in 3D-printed components produced via Fused Deposition Modeling (FDM) using polycarbonate (PC) and polylactic acid (PLA). Through a systematic full factorial experimental design, three critical parameters were examined: material types (PC and PLA), layer thickness (0.2 mm and 0.4 mm), and build orientation (horizontal and vertical). Preliminary trends suggest that vertically oriented specimens showed up to 64.7% higher tensile strength compared to horizontal builds, though with significantly reduced ductility. Contributing to growing evidence regarding layer thickness effects, thicker layers (0.4 mm) showed improved ultimate strength by up to 36.2% while simultaneously reducing production time by 50%. However, statistical power analysis revealed insufficient sample size (n = 1 per condition) to establish significance for orientation effects, despite large practical differences observed. PC specimens demonstrated superior strength (maximum 67.5 MPa) and fracture energy, while PLA offered better ductility (up to 22.4% strain). These exploratory findings provide promising directions for future adequately powered investigations for tailored parameter selection according to specific application requirements. Full article

In this paper, a target detection network algorithm based on a multi-branch convolution and attention improvement Cross-Stage Partial-Fusion Bottleneck with Two Convolutions (C2F) module is proposed for the difficult task of detecting small targets in unmanned aerial vehicles. A C2F module method consisting of fusing partial convolutional (PConv) layers was designed to improve the speed and efficiency of extracting features, and a method consisting of combining multi-scale feature fusion with a channel space attention mechanism was applied in the neck network. An FA-Block module was designed to improve feature fusion and attention to small targets’ features; this design increases the size of the miniscule target layer, allowing richer feature information about the small targets to be retained. Finally, the lightweight up-sampling operator Content-Aware ReAssembly of Features was used to replace the original up-sampling method to expand the network’s sensory field. Experimental tests were conducted on a self-complied mountain pedestrian dataset and the public VisDrone dataset. Compared with the base algorithm, the improved algorithm improved the mAP50, mAP50-95, P-value, and R-value by 2.8%, 3.5%, 2.3%, and 0.2%, respectively, on the Mountain Pedestrian dataset and the mAP50, mAP50-95, P-value, and R-value by 9.2%, 6.4%, 7.7%, and 7.6%, respectively, on the VisDrone dataset. Full article

In the context of the new power system, high-current switchgear is prone to various faults due to complex operation environments and severe load fluctuations. Among them, an abnormal temperature rise can lead to contact oxidation, insulation aging, and even equipment failure, posing a serious threat to the safety of the distribution system. The operation risk assessment of high-current switchgear has thus become a key to ensuring the safety of the distribution system. Ensemble learning, which integrates the advantages of multiple models, provides an effective approach for accurate and intelligent risk assessment. However, existing ensemble learning methods have shortcomings in feature extraction, time-series modeling, and generalization ability. Therefore, this paper first preprocesses and reduces the dimensionality of multi-source data, such as historical load and equipment operation status. Secondly, we propose an operation risk assessment method for high-current switchgear based on ensemble learning: in the first layer, an improved random forest (RF) is used to optimize feature extraction; in the second layer, an improved long short-term memory (LSTM) network with an attention mechanism is adopted to capture time-series dependent features; in the third layer, an adaptive back propagation neural network (ABPNN) model fused with an adaptive genetic algorithm is utilized to correct the previous results, improving the stability of the assessment. Simulation results show that in temperature rise prediction, the proposed algorithm significantly improves the goodness-of-fit indicator with increases of 15.4%, 4.9%, and 24.8% compared to three baseline algorithms, respectively. It can accurately assess the operation risk of switchgear, providing technical support for intelligent equipment operation and maintenance, and safe operation of the system. Full article

Thermoset fibre-reinforced composites are widely used in high-end industries, but a growing demand for more sustainable and recyclable alternatives conveyed the research efforts towards thermoplastics. To expand their usage, new approaches to their manufacture and mechanical performance must be tackled and tailored to each engineering challenge. The present study designed, manufactured and tested advanced multi-layer laminated composites of thermoplastic polypropylene prepreg reinforced with continuous woven fibreglass with interlayer toughening through thermoplastic polyurethane elastomer (TPU) layers manufactured by fused filament fabrication. The manufacturing process was iteratively optimized, resulting in successful adhesion between layers. Three composite configurations were produced: baseline glass fibre polypropylene (GFPP) prepreg and two multi-layer composites, with solid and honeycomb structured TPU layers. Thermal and mechanical analyses were conducted with both the polyurethane elastomer and the manufactured laminates. Tensile testing was conducted on additively manufactured polyurethane elastomer specimens, while laminated composites were tested in three-point bending. The results demonstrated the potential of the developed laminates. TPU multi-layer laminates exhibit higher thermal stability compared to the baseline GFPP prepreg-based composites. The addition of elastomeric layers decreases the flexural modulus but increases the ability to sustain plastic deformation. Multi-layer laminate composites presenting honeycomb TPU layers exhibit improved geometric and mechanical consistency, lower delamination and fibre breakage, and a high elastic recoverability after testing. Full article

Additive manufacturing involves numerous independent parameters, often leading to inconsistent print quality and necessitating costly trial-and-error approaches to optimize input variables. Machine learning offers a solution to this non-linear problem by predicting optimal printing parameters from a minimal set of experiments. Using Fused Deposition Modeling (FDM) as a case study, this work develops a machine learning-powered process to predict porosity defects. Specimens in two geometrical scales were 3D-printed and CT-scanned, yielding raw datasets of grayscale images. A machine learning image classifier was trained on the small-cube dataset (~2200 images) to distinguish exploitable images from defective ones, averaging over 97% accuracy and correctly classifying more than 90% of the large-cube exploitable images. The developed preprocessing scripts extracted porosity features from the exploitable images. A repeatability study analyzed three replicate specimens printed under identical conditions, and quantified the intrinsic process variability, showing an average porosity standard deviation of 0.47% and defining an uncertainty zone for quality control. A multi-layer perceptron (MLP) was independently trained on 1709 data points derived from the small-cube dataset and 3746 data points derived from the large-cube dataset. Its accuracy was 54.4% for the small cube and increased to 77.6% with the large-cube dataset, due to the larger sample size. A rigorous grouped k-fold cross-validation protocol, relying on splitting data per cube, strengthened the ML algorithms against data leakage and overfitting. Finally, a dimensional scalability study further assessed the use of the pipeline for the large-cube dataset and established the impact of geometrical scaling on defect formation and prediction in 3D-printed parts. Full article

Traffic flow prediction is a prominent research area in intelligent transportation systems, significantly contributing to urban traffic management and control. Existing methods or models for traffic flow prediction predominantly rely on a fixed-graph structure to capture spatial correlations within a road network. However, the fixed-graph structure can restrict the representation of spatial information due to varying conditions such as time and road changes. Drawing inspiration from the attention mechanism, a new prediction model based on the mixed-graph neural network is proposed to dynamically capture the spatial traffic flow correlations. This model uses graph convolution and attention networks to adapt to complex and changeable traffic and other conditions by learning the static and dynamic spatial traffic flow characteristics, respectively. Then, their outputs are fused by the gating mechanism to learn the spatial traffic flow correlations. The Transformer encoder layer is subsequently employed to model the learned spatial characteristics and capture the temporal traffic flow correlations. Evaluated on five real traffic flow datasets, the proposed model outperforms the state-of-the-art models in prediction accuracy. Furthermore, ablation experiments demonstrate the strong performance of the proposed model in long-term traffic flow prediction. Full article

Optimizing printing parameters is crucial for enhancing the efficiency, surface quality, and dimensional accuracy of Fused Deposition Modeling (FDM) processes. A review of numerous publications reveals that most scholars analyze factors such as nozzle diameter and printing speed, while few investigate the impact of layer thickness, infill density, and shell layer count on print quality. Therefore, this study employed 3D slicing software to process the three-dimensional model and design printing process parameters. It systematically investigated the effects of layer thickness, infill density, and number of shells on printing time and geometric accuracy, quantifying the evaluation through volumetric error. Using an ABS connecting rod model, optimal parameters were determined within the defined range through orthogonal experimental design and signal-to-noise ratio (S/N) analysis. Subsequently, a backpropagation (BP) neural network was constructed to establish a predictive model for process optimization. Results indicate that parameter selection significantly impacts print duration and surface quality. Validation confirmed that the combination of 0.1 mm layer thickness, 40% infill density, and 5-layer shell configuration achieves the highest dimensional accuracy (minimum volumetric error and S/N value). Under this configuration, the volumetric error rate was 3.062%, with an S/N value of −9.719. Compared to other parameter combinations, this setup significantly reduced volumetric error, enhanced surface texture, and improved overall print precision. Statistical analysis indicates that the BP neural network model achieves a Mean Absolute Percentage Error (MAPE) of no more than 5.41% for volume error rate prediction and a MAPE of 5.58% for signal-to-noise ratio prediction. This validates the model’s high-precision predictive capability, with the established prediction model providing effective data support for FDM parameter optimization. Full article

With the rapid development of modern industry, people’s living pressures are gradually increasing, and an increasing number of individuals are affected by sleep disorders such as insomnia, hypersomnia, and sleep apnea syndrome. Many cardiovascular and psychiatric diseases are also closely related to sleep. Therefore, the early detection, accurate diagnosis, and treatment of sleep disorders an urgent research priority. Traditional manual sleep staging methods have many problems, such as being time-consuming and cumbersome, relying on expert experience, or being subjective. To address these issues, researchers have proposed multiple algorithmic strategies for sleep staging automation based on deep learning in recent years. This paper studies MASleepNet, a sleep staging neural network model that integrates multimodal deep features. This model takes multi-channel Polysomnography (PSG) signals (including EEG (Fpz-Cz, Pz-Oz), EOG, and EMG) as input and employs a multi-scale convolutional module to extract features at different time scales in parallel. It then adaptively weights and fuses the features from each modality using a channel-wise attention mechanism. The integrated temporal features are integrated into a Bidirectional Long Short-Term Memory (BiLSTM) sequence encoder, where an attention mechanism is introduced to identify key temporal segments. The final classification result is produced by the fully connected layer. The proposed model was experimentally evaluated on the Sleep-EDF dataset (consisting of two subsets, Sleep-EDF-78 and Sleep-EDF-20), achieving classification accuracies of 82.56% and 84.53% on the two subsets, respectively. These results demonstrate that deep models that integrate multimodal signals and an attention mechanism offer the possibility to enhance the efficiency of automatic sleep staging compared to cutting-edge methods. Full article

The polymer deposition layer (PDL) formed during inductively coupled plasma (ICP) processing significantly limits the figuring accuracy and surface quality of fused silica optics. This study investigates the formation mechanism, composition, and evolution of the PDL under varying dwell times and proposes an innovative dwell time gradient strategy to suppress roughness deterioration. A significant disparity in hardness and elastic modulus between the deposition layer and the substrate is revealed, explaining its preferential removal and protective buffering effect in computer-controlled optical surfacing (CCOS). A hybrid ICP-CCOS polishing process was developed for processing a ϕ100 mm fused silica mirror. The results show that within 33 min, the surface graphic error RMS was significantly reduced from 58.006 nm to 12.111 nm, and within 90 min, the surface roughness was ultra-precisely reduced from Ra 1.719 nm to Ra 0.151 nm. The average processing efficiency was approximately 0.63 cm²/min. Critically, a damage-free, ultra-smooth surface without subsurface damage (SSD) was successfully achieved. This hybrid process enables the simultaneous optimization of figure accuracy and roughness, eliminating the need for iterative figuring cycles. It provides a novel theoretical framework for high-precision figuring and post-ICP polymer removal, advancing the efficient fabrication of high-performance optics. Full article

With the advancement of image processing techniques, few-shot learning (FSL) has gradually become a key approach to addressing the problem of data scarcity. However, existing FSL methods often rely on unimodal information under limited sample conditions, making it difficult to capture fine-grained differences between categories. To address this issue, we propose a multimodal few-shot learning method based on category name expansion and image feature enhancement. By integrating the expanded category text with image features, the proposed method enriches the semantic representation of categories and enhances the model’s sensitivity to detailed features. To further improve the quality of cross-modal information transfer, we introduce a cross-modal residual connection strategy that aligns features across layers through progressive fusion. This approach enables the fused representations to maximize mutual information while reducing redundancy, effectively alleviating the information bottleneck caused by uneven entropy distribution between modalities and enhancing the model’s generalization ability. Experimental results demonstrate that our method achieves superior performance on both natural image datasets (CIFAR-FS and FC100) and a medical image dataset. Full article

Brain tumors usually appear as masses formed by localized abnormal cell proliferation. Although complete removal of tumors is an ideal treatment goal, this process faces many challenges due to the aggressive nature of malignant tumors and the need to protect normal brain tissue. Therefore, early diagnosis is crucial to mitigate the harm posed by brain tumors. In this study, the classification accuracy is improved by improving the ResNet50 model. Specifically, the image is preprocessed and enhanced firstly, and the image is denoised by fractional calculus; then, transfer learning technology is adopted, the ECA attention mechanism is introduced, the convolutional layer in the residual block is optimized, and the multi-scale convolutional layer is fused. These optimization measures not only enhance the model’s ability to grasp the overall details but also improve its ability to recognize micro and macro features. This allows the model to understand data features more comprehensively and process image details more efficiently, thereby improving processing accuracy. In addition, the improved ResNet50 model is combined with EfficientNetB0 to further optimize performance and improve classification accuracy by utilizing EfficientNetB0’s efficient feature extraction capabilities through feature fusion. In this study, we used a brain tumor image dataset containing 5712 training images and 1311 validation images. The optimized ResNet50 model achieves a verification accuracy of 98.78%, which is 3.51% higher than the original model, and the Kappa value is also increased by 4.7%. At the same time, the lightweight design of the EfficientNetB0 improves performance while reducing uptime. These improvements can help diagnose brain tumors earlier and more accurately, thereby improving patient outcomes and survival rates. Full article

Point cloud completion for cotton leaves is critical for accurately reconstructing complete shapes from sparse and significantly incomplete data. Traditional methods typically assume small missing ratios (≤25%), which limits their effectiveness for morphologically complex cotton leaves with severe sparsity (50–75%), large geometric distortions, and extensive point loss. To overcome these challenges, we introduce an end-to-end neural network that combines PF-Net and PointNet++ to effectively reconstruct dense, uniform point clouds from incomplete inputs. The model initially uses a multiresolution encoder to extract multiscale features from locally incomplete point clouds at different resolutions. By capturing both low-level and high-level attributes, these features significantly enhance the network’s ability to represent semantic content and geometric structure. Next, a point pyramid decoder generates missing point clouds hierarchically from layers at different depths, effectively reconstructing the fine details of the original structure. PointNet++ is then used to fuse and reshape the incomplete input point clouds with the generated missing points, yielding a fully reconstructed and uniformly distributed point cloud. To ensure effective task completion at different training stages, a loss function freezing strategy is employed, optimizing the network’s performance throughout the training process. Experimental evaluation on the cotton leaf dataset demonstrated that the proposed model outperformed PF-Net, reducing the Chamfer distance by 80.15% and the Earth Mover distance by 54.35%. These improvements underscore the model’s robustness in reconstructing sparse point clouds for precise agricultural phenotyping. Full article

Underwater images frequently exhibit color distortion, detail blurring, and contrast degradation due to absorption and scattering by the underwater medium. This study proposes a progressive color correction strategy integrated with a vision-inspired image enhancement framework to address these issues. Specifically, the progressive color correction process includes adaptive color quantization-based global color correction, followed by guided filter-based local color refinement, aiming to restore accurate colors while enhancing visual perception. Within the vision-inspired enhancement framework, the color-adjusted image is first decomposed into a base layer and a detail layer, corresponding to low- and high-frequency visual information, respectively. Subsequently, detail enhancement and noise suppression are applied in the detail pathway, while global brightness correction is performed in the structural pathway. Finally, results from both pathways are fused to yield the enhanced underwater image. Extensive experiments on four datasets verify that the proposed method effectively handles the aforementioned underwater enhancement challenges and significantly outperforms state-of-the-art techniques. Full article

Accurate human keypoint localization in complex environments demands robust sensing and advanced modeling. In this article, we construct a ST-ConvLSTM network for 3D human keypoint estimation via millimeter-wave radar point clouds. The ST-ConvLSTM network processes multi-channel radar image inputs, generated from multi-frame fused point clouds through parallel pathways. These pathways are engineered to extract rich spatiotemporal features from the sequential radar data. The extracted features are then fused and fed into fully connected layers for direct regression of 3D human keypoint coordinates. In order to achieve better network performance, a mmWave radar 3D human keypoint dataset (MRHKD) is built with a hybrid human motion annotation system (HMAS), in which a binocular camera is used to measure the human keypoint coordinates and a 60 GHz 4T4R radar is used to generate radar point clouds. Experimental results demonstrate that the proposed ST-ConvLSTM, leveraging its unique ability to model temporal dependencies and spatial patterns in radar imagery, achieves MAEs of 0.1075 m, 0.0633 m, and 0.1180 m in the horizontal, vertical, and depth directions. This significant improvement underscores the model’s enhanced posture recognition accuracy and keypoint localization capability in challenging conditions. Full article