Hybrid Network Structure Research Articles

Light field images super-resolution method based on hybrid low-dimensional spatial–angular interaction feature and linear complementation epipolar feature

Light-field (LF) cameras record 4D information about the intensity and angle of light, but limited by the resolution of the sensor, LF images require a trade-off between spatial and angular resolution, which results in generally low spatial resolution for LF images. We work on spatial super-resolution reconstruction (SR) of LF images, but due to the high dimensionality of 4D LF, it is difficult to process it directly, so we decompose 4D LF into low-dimensional sub-spaces. Since the 4D features of interest to different low-dimensional sub-spaces are different, we design a hybrid network structure—HSAESR, which processes multiple low-dimensional sub-spaces individually to adequately model the texture information in the 4D LF. HSAESR is designed to reconstruct Macro Pixel Images (MacPI) and Extremely Planar Images (EPI) respectively, and the final reconstruction results are weighted and fused. For MacPI, a CNN-based feature interaction module is designed using a novel asymptotic ’grid’ interaction approach to fully utilize the correlation between the 2D spatial dimension and the 2D angular dimension sub-spaces in 4D LF. For EPI, a Transformer-based epipolar feature linear complementation module is designed learning global and non-global multi-directional epipolar feature correlations, which learns feature linear correlation factor (α) and linear bias weight (β) through the correlation of non-global epipolar features, and corrects and complements the global epipolar correlation features through feature linear complementation. Extensive experiments are conducted on five datasets, the results show that HSAESR further improves the performance of LF images spatial SR, the reconstruction results are better than the comparison methods.

An Active Learning Semantic Segmentation Model Based on an Improved Double Deep Q-Network

Image semantic segmentation is essential in fields such as computer vision, autonomous driving, and human-computer interaction due to its ability to accurately identify and classify each pixel in an image. However, this task is fraught with challenges, including the difficulty of obtaining detailed pixel labels and the problem of class imbalance in segmentation datasets. These challenges can hinder the effectiveness and efficiency of segmentation models. To address these issues, we propose an active learning semantic segmentation model named CG_D3QN, which is designed and implemented based on an enhanced Double Deep Q-Network (D3QN). The proposed CG_D3QN model incorporates a hybrid network structure that combines a dueling network architecture with Gated Recurrent Units (GRUs). This novel approach improves policy evaluation accuracy and computational efficiency by mitigating a Q-value overestimation and making better use of historical state information. Our experiments, conducted on the CamVid and Cityscapes datasets, reveal that the CG_D3QN model significantly reduces the number of required sample annotations by 65.0% compared to traditional methods. Additionally, it enhances the mean Intersection over Union (IoU) for underrepresented categories by approximately 1% to 3%. These results highlight the model’s effectiveness in lowering annotation costs, addressing class imbalance, and its versatility across different segmentation networks.

Enhancing cement-based composites via regulated hydration and concurrent construction of robust organic-inorganic network

Cement-based materials with excellent mechanical properties are in high demand in the civil engineering. This work utilized an innovative in-situ copolymerization approach for synergistic modification of cement-based materials. Vinyl hybrid silica microspheres (VSM) were synthesized through the sol-gel method. Subsequently, With the assistance of initiators and cross-linking agents, the VSM and acrylamide (AM) in-situ copolymerization and were uniformly dispersed in the cement matrix, leading to the successful development of a modified cement-based material. The copolymerization of VSM and AM resulted in the creation of a comprehensive three-dimensional hybrid network structure, interconnected by covalent bonds between the constituent components. Experimental results unequivocally demonstrate that the mechanical properties of the composite have undergone remarkable enhancement. Notably, the flexural strength experienced a substantial increase due to the in-situ polymerization of AM. Furthermore, the compressive strength was significantly improved, achieving a 21.4 % increase compared to samples modified with 2 wt% AM alone. The enhancement can be attributed to the synergistic effect of the in-situ copolymerization structure and the VSM attached to it. The microspheres not only aided in the development of a robust cement matrix and a higher compactness, but also bolstered the compressive modulus of the flexible polymer chains. Collectively, these factors contributed to an increase in the overall resistance to deformation, comprehensively enhancing the mechanical performance of the material. This study addresses the challenges posed by the loss of compressive strength during in-situ polymerization modification, introduces an innovative methodology aimed at significantly enhancing the mechanical properties of cement-based materials.

An aspect sentiment analysis model with Aspect Gated Convolution and Dual-Feature Filtering layers

Aspect level sentiment analysis is a basic task to determine the sentiment bias based on the contextual information near the aspect words. Some sentences contain many confusing feature words due to incomplete structure or high complexity in sentiment prediction, which can easily lead to the problem that the model pays too much attention to unimportant features. Furthermore, the role of aspect-related sentiment features is not significant in the feature extraction process. The capsule network, due to its complex network structure, leads to capsule detachment when too much information is found in the dataset. Too much computational resources are easily consumed during dynamic routing, which in turn affects the model's judgement of the polarity of aspect-related sentiment. To address these issues, we propose a capsule network (AGCDFF-Caps) with aspect-gated convolution and Dual-Feature Filtering layers for aspect-level sentiment analysis. A pre-trained BERT model is used to generate a serialised representation of the text, and the semantic representation in the contextual text is enhanced by the self-attention mechanism and bi-GRU. An aspect-gating based convolutional neural network is constructed to selectively extract contextual sentiment information about aspects and discard irrelevant information. A capsule network is then used to learn the multispatial semantic features of the aspect and the text. In particular, we incorporate a Dual-Feature Filtering network structure into the capsule network structure to strengthen the interaction between the particular aspect and the context from global and local perspectives, filtering the redundant information in local semantics and global semantics. The opinion feature representations that can more accurately express the emotional tendency of the aspect are obtained. Experimental results on SemEval2014 and Twitter datasets show that the proposed hybrid network structure has superior classification performance compared to 12 advanced baseline methods.

Porosity prediction using a deep learning method based on bidirectional spatio-temporal neural network

Deep learning is one of the best machine learning algorithms for modeling complex mapping relationships between independent and dependent variables, and thus it can be viewed as an ideal approach to predict porosity. In this study, to overcome the deficiencies in current porosity prediction based on deep learning and improve the prediction accuracy, we proposed a deep learning model based on bidirectional temporal convolutional network (BTCN) and bidirectional long short-term memory (BLSTM) network, called bidirectional spatio-temporal neural network (BSTNN), to establish a porosity prediction model. First, the maximum information coefficient is used to analyze the correlation between well logs and porosity, which provides a basis for determining the inputs of the prediction model. Then, a hybrid network structure is constructed by using BTCN and BLSTM, in which BTCN goes to learn the bidirectional long sequence features and BLSTM goes to learn the variation trend and context information with depth, so the hybrid network structure can learn richer logging signal features. Finally, the extracted features are passed through the fully connected layer to output the porosity prediction results. Porosity prediction experiment are conducted by using the actual field data set. The results show that the proposed method has the lower prediction errors for the porosity modeling (RMSE = 0.368 and MAE = 0.260) compared to the benchmark models convolutional neural network (RMSE = 0.404 and MAE = 0.292) and long short-term memory network (RMSE = 0.418 and MAE = 0.298), which verifies the effectiveness of this prediction method.

Maritime vessel classification based on a dual network combining EfficientNet with a hybrid network MPANet

AbstractShip classification is an important technique for enhancing maritime management and security. Visible and infrared sensors are generally employed to deal with the challenging problem and improve classification performance. Herein, a two‐branch feature fusion neural network structure is proposed to classify the visible and infrared maritime vessel images simultaneously. Specifically, in this two‐branch neural network, one branch is based on a deep convolutional neural network that is used to extract the visible image features, while the other is a hybrid network structure that is a multi‐scale patch embedding network called MPANet. The sub‐network MPANet can extract fine‐ and coarse‐grained features, in which the pooling operation instead of the multi‐head attention mechanism is utilized to reduce memory consumption. When there are infrared images, it is used to extract the infrared image features, otherwise, this branch is also utilized to extract visible image features. Therefore, this dual network is suitable with or without infrared images. The experimental results on the visible and infrared spectrums (VAIS) dataset demonstrate that the introduced network achieves state‐of‐the‐art ship classification performance on visible images and paired visible and infrared ship images.

Enhanced thermal conductivity of PVDF by filling with T‐ZnOw and MWCNT based on hybrid 3D network

AbstractIn this work, a novel approach is taken to enhance the thermal conductivity and mechanical properties of composites by developing highly efficient thermally conductive PVDF‐based composites. This is developed by incorporating tetra‐pod zinc oxide whiskers (T‐ZnOw) in conjunction with multiwalled carbon nanotubes (MWCNT) to form binary hybrid fillers within a polyvinylidene fluoride (PVDF) matrix. Polydopamine (PDA), characterized by physical adsorption, is selected for surface coating on T‐ZnOw since T‐ZnOw is easy to self‐agglomerate and difficult to modify by physical and chemical methods. The coating aims to improve the dispersibility of T‐ZnOw and reduce the interfacial thermal resistance with the substrate. MWCNT is oxidized to improve its dispersion ability while strengthening the interaction with the matrix, due to the absence of interaction between MWCNTs and polymer matrix and poor dispersibility with the matrix. The dispersity of PDA@T‐ZnOw and o‐MWCNT is improved, and the three‐dimensional hybrid network structure can be better constructed to improve the thermal conductivity of the composite membrane. At a filler content of 25 wt%, the thermal conductivity reaches 0.302 W m−1 K−1 for T‐ZnOw–MWCNT–PVDF and 0.222 W m−1 K−1 for PDA@T‐ZnOw–o‐MWCNT–PVDF.

Rheological study of hybrid aqueous suspension of TEMPO-oxidized cellulose nanofibrils and graphene oxide

The incorporation of graphene oxide (GO) into a nanocellulose matrix has attracted considerable attention due to the unique advantages of both components. This study focuses on investigating the viscoelastic and flow properties of hybrid aqueous suspensions (2.00 w/v%), composed of TEMPO-oxidized cellulose nanofibrils (TOCNF) and GO at different TOCNF/GO weight ratios. To adjust the elastic properties of the hybrid suspensions, calcium ions are introduced, varying the concentration systematically to study their effects on the hybrid network structure. All blends exhibit shear-thinning behaviour and demonstrate elastic, gel-like properties. Notably, in the absence of calcium ions, the enhancement of elastic properties is more pronounced at higher GO fractions. Conversely, with the introduction of calcium ions, the enhancement of elastic properties becomes particularly important at higher TOCNF fractions. For the quantitative evaluation of these enhancements, we employ the logarithmic mixing rule. Significant positive deviations from the predictions of the logarithmic mixing rule are ascribed to the complex, concentration-dependent arrangement of cellulose nanofibrils and GO liquid crystals in the aqueous suspension, coupled with ionic crosslinking induced by calcium ions. The study aims to contribute to the understanding of the rheological behaviour of the TOCNF/GO hydrogel, showing potential advancements in various applications.Graphical abstract

Non-Terrestrial Network Concepts in 5G and Beyond Communication Technologies

Mobile communication systems used in 5G and beyond will provide high-capacity, reliable and low-latency services within wide coverage areas. Currently, terrestrial networks are widely used, but it is expected that these networks will be replaced by hybrid network solutions where terrestrial and non-terrestrial networks operate together. These hybrid ones can perform cutting-edge complimentary solutions when the traditional networks have weak signal levels and geographical difficulties. The main motivation of this study can be stated as the increasing use of hybrid network structures for the solution of digital divine and access problems. In this study, design optimizations of hybrid wireless networks according to different system architectures and performance analyzes are given and various non-terrestrial network technologies are compared.

MHDNet: A Multi-Scale Hybrid Deep Learning Model for Person Re-Identification

The primary objective of person re-identification is to identify individuals from surveillance videos across various scenarios. Conventional pedestrian recognition models typically employ convolutional neural network (CNN) and vision transformer (ViT) networks to extract features, and while CNNs are adept at extracting local features through convolution operations, capturing global information can be challenging, especially when dealing with high-resolution images. In contrast, ViT rely on cascaded self-attention modules to capture long-range feature dependencies, sacrificing local feature details. In light of these limitations, this paper presents the MHDNet, a hybrid network structure for pedestrian recognition that combines convolutional operations and self-attention mechanisms to enhance representation learning. The MHDNet is built around the Feature Fusion Module (FFM), which harmonizes global and local features at different resolutions. With a parallel structure, the MHDNet model maximizes the preservation of local features and global representations. Experiments on two person re-identification datasets demonstrate the superiority of the MHDNet over other state-of-the-art methods.

Construction of micro-nano hybrid structure based on carbon nanotube whisker and alumina for thermally conductive yet electrically insulating silicone rubber composites

High-performance electronics urgently need more effective thermally conductive rubber composites to solve interfacial heat transfer problems in the thermal management systems. Tiny amounts nanocarbon materials (NCM) can significantly improve the thermal conductivity of conventional ceramic-filled rubber composites, but the volume exclusion effect of micrometer ceramic fillers makes NCM highly susceptible to the formation of the conductive pathways, which inevitably leads to the substantial decrease in the volume resistivity of the materials, posing a safety hazard, such as short circuits, to electronic devices. Here, we report an electrostatic self-assembly method to prepare CNW@n-Al2O3 hybrids by loading nano-alumina (n-Al2O3) onto carbon nanotube whiskers (CNW) and co-filling them with micrometer alumina (m-Al2O3) to silicone rubber, constructing a micro-nano-multi-level hybrid network structure, which can fully utilize the high thermal conductivity while shielding the electrical conductivity of CNW. The resulting composite filled with 2 phr of CNW@n-Al2O3 exhibits a significantly enhanced thermal conductivity of 1.137 W/(m·K) and a high volume resistance of 1.323 × 109 Ω cm, and is proved to be used as an excellent thermal interface material to assist the heat dissipation of the microelectronic chip. This study provides a facile and effective strategy for the design of thermally conductive yet electrically insulating rubber composites filled with CNW, which shows a bright application prospect in the thermal management of high-performance electronic devices.

Hierarchical carbon nanotubes-modified heterogeneous composites derived from melamine-mixed ZIF-67/MXene for broadband microwave absorption

In this paper, carbon nanotubes (CNTs) modified TiO2/C/Co@C composites (TiO2/C/Co@C(CNTs)) with good electromagnetic (EM) wave properties were synthesized originated from ZIF-67/MXene composites supplemented with external carbon source melamine by in situ synthesis and high temperature pyrolysis. The effect of the melamine content on the microstructure, EM parameters, and the microwave absorption (MWA) properties of the TiO2/C/Co@C(CNTs) composites was investigated. The results showed that the in-situ grown CNTs were organically connected to the pyrolyzed ZIF-67/MXene derivatives to form a hybrid network structure. The composites achieve the minimum reflection loss (RLmin) value of −47.52 dB at a thickness of 1.5 mm, the effective absorption bandwidth (EAB) of 5.06 GHz at 1.8 mm with the RLmin value of −41.35 dB, and the RLmin value of −52.43 dB at 6.18 GHz. The excellent EM wave attenuation ability of the TiO2/C/Co@C(CNTs) composites covers nearly the entire frequency band of 2–18 GHz with the RLmin value almost less than −30 dB in the thickness of 1.5–5.0 mm. The unique microstructure, dielectric-magnetic synergistic effect, and multiple interfacial polarization optimize the impedance matching. This word proposes a promising new strategy for constructing highly efficient MOF-derived and MXene-based microwave absorbers.

DCTNET: HYBRID NETWORK MODEL FUSING WITH MULTISCALE DEFORMABLE CNN AND TRANSFORMER STRUCTURE FOR ROAD EXTRACTION FROM GAOFEN SATELLITE REMOTE SENSING IMAGE

Abstract. The urban road network detection and extraction have significant applications in many domains, such as intelligent transportation and navigation, urban planning, and automatic driving. Although manual annotation methods can provide accurate road network maps, their low efficiency with high-cost consumption are insufficient for the current tasks. Traditional methods based on spectral or geometric information rely on shallow features and often struggle with low semantic segmentation accuracy in complex remote sensing backgrounds. In recent years, deep convolutional neural networks (CNN) have provided robust feature representations to distinguish complex terrain objects. However, these CNNs ignore the fusion of global-local contexts and are often confused with other types of features, especially buildings. In addition, conventional convolution operations use a fixed template paradigm to aggregate local feature information. The road features present complex linear-shape geometric relationships, which brings some obstacles to feature construction. To address the above issues, we proposed a hybrid network structure that combines the advantages of CNN and transformer models. Specifically, a multiscale deformable convolution module has been developed to capture local road context information adaptively. The Transformer model is introduced into the encoder to enhance semantic information to build the global context. Meanwhile, the CNN features are fused with the transformer features. Finally, the model outputs a road extraction prediction map in high spatial resolution. Quantitative analysis and visual expression confirm that the proposed model can effectively and automatically extract road features from complex remote sensing backgrounds, outperforming state-of-the-art methods with IOU by 86.5% and OA by 97.4%.

Effect of blending ratio on the compatibility and stability of konjac glucomannan/starch composite systems

A series of complex dispersions and films combining Konjac glucomannan (KGM) and starch in different ratios were prepared. The compatibility and stability of KGM/Starch composite (KS) systems were evaluated using a new horizontal casting device coupled with CCD camera system and Near-infrared (NIR) spectrometer and multiple conventional characterization techniques. The findings from the casting device revealed that when the KGM and starch were combined in a 7∶3 ratio, the dispersion demonstrated isotropic behavior, flowing and diffusing at a consistent speed from the central point to the surrounding area on the casting disc. Additionally, the scattered particles in the NIR reflectance spectrum were evenly distributed at every point, indicating excellent compatibility and uniformity within the dispersion. FTIR and XRD analyses indicated the existence of molecular interactions between the film components, and showed good compatibility between KGM and starch when the KGM content was greater than 50%. The best compatibility was obtained when the KGM content was 70%. The SEM results showed that the increase of KGM content contributed to a smoother and denser microstructure in the cross-section of the KS composite films, especially when the KGM content was 70% (w/w). The rheological results of the KS dispersion further confirmed the formation of a hybrid network structure between KGM and starch. Higher KGM content led to more hydrogen bonds and provided more opportunities for macromolecular chain entanglement. This work will contribute to a better understanding of the interaction mechanism between polysaccharides and starch, as well as provide a new perspective and novel characterization methods to deconstruct the compatibility and assembly structure stability between KGM and starch. It is beneficial for facilitate the application of KGM in the development of denatured food or starch materials, including food texture properties and emulsion stability, and establish a foundation for the diversification of food design tools.

Cformer: An underwater image enhancement hybrid network combining convolution and transformer

AbstractUnderwater images are the most direct and effective ways to obtain underwater information. However, underwater images typically suffer from contrast reduction and colour distortion due to the absorption and scattering of water by light, which seriously limits the further development of underwater visual tasks. Recently, the convolutional neural network has been extensively applied in underwater image enhancement for its powerful local information extraction capabilities, but due to the locality of convolution operation, it cannot capture the global context well. Although the recently emerging Transformer can capture global context, it cannot model local correlations. Cformer is proposed, which is an Unet‐like hybrid network structure. First, a Depth Self‐Calibrated block is proposed to extract the local features of the image effectively. Second, a novel Cross‐Shaped Enhanced Window Transformer block is proposed. It captures long‐range pixel interactions while dramatically reducing the computational complexity of feature maps. Finally, the depth self‐calibrated block and the cross‐shaped enhanced window Transformer block are ingeniously fused to build a global–local Transformer module. Extensive ablation studies are performed on public underwater datasets to demonstrate the effectiveness of individual components in the network. The qualitative and quantitative comparisons indicate that Cformer achieves superior performance compared to other competitive models.

Conformer: Local Features Coupling Global Representations for Recognition and Detection.

With convolution operations, Convolutional Neural Networks (CNNs) are good at extracting local features but experience difficulty to capture global representations. With cascaded self-attention modules, vision transformers can capture long-distance feature dependencies but unfortunately deteriorate local feature details. In this paper, we propose a hybrid network structure, termed Conformer, to take both advantages of convolution operations and self-attention mechanisms for enhanced representation learning. Conformer roots in feature coupling of CNN local features and transformer global representations under different resolutions in an interactive fashion. Conformer adopts a dual structure so that local details and global dependencies are retained to the maximum extent. We also propose a Conformer-based detector (ConformerDet), which learns to predict and refine object proposals, by performing region-level feature coupling in an augmented cross-attention fashion. Experiments on ImageNet and MS COCO datasets validate Conformer's superiority for visual recognition and object detection, demonstrating its potential to be a general backbone network. Code is available at https://github.com/pengzhiliang/Conformer.

Bottom hole pressure prediction based on hybrid neural networks and Bayesian optimization

Many scholars have focused on applying machine learning models in bottom hole pressure (BHP) prediction. However, the complex and uncertain conditions in deep wells make it difficult to capture spatial and temporal correlations of measurement while drilling (MWD) data with traditional intelligent models. In this work, we develop a novel hybrid neural network, which integrates the Convolution Neural Network (CNN) and the Gate Recurrent Unit (GRU) for predicting BHP fluctuations more accurately. The CNN structure is used to analyze spatial local dependency patterns and the GRU structure is used to discover depth variation trends of MWD data. To further improve the prediction accuracy, we explore two types of GRU-based structure: skip-GRU and attention-GRU, which can capture more long-term potential periodic correlation in drilling data. Then, the different model structures tuned by the Bayesian optimization (BO) algorithm are compared and analyzed. Results indicate that the hybrid models can extract spatial-temporal information of data effectively and predict more accurately than random forests, extreme gradient boosting, back propagation neural network, CNN and GRU. The CNN-attention-GRU model with BO algorithm shows great superiority in prediction accuracy and robustness due to the hybrid network structure and attention mechanism, having the lowest mean absolute percentage error of 0.025%. This study provides a reference for solving the problem of extracting spatial and temporal characteristics and guidance for managed pressure drilling in complex formations.

Epiphyte-inspired multifunctional biocomposites for electromagnetic interference shielding

With the development of advanced communication systems, paralleled with recent efforts to adopt lightweight, low-carbon footprint materials, there is an urgent demand for green electromagnetic interference (EMI) shielding materials. However, this is challenged by the shielding effectiveness of bio-based EMI and their capacity to eliminate radiation pollution while meeting the demands of diverse application environments. Learning from the strategies used by epiphytes, we introduce a highly efficient and multi-functional bio-based EMI shielding eco-material. We developed 3D hybrid network structures incorporating coordination-driven silver nanoparticles to achieve excellent shielding effectiveness, ∼30.8 dB, while simultaneously meeting mechanical demands, multi-functionality, water resistance, flame retardancy, and antibacterial properties. Overall, this work provides a novel but practical pathway to implement bio-based materials conferred with EMI protection, applicable in next-generation electronics.

Phase formation and mechanical properties of graphene reinforced regolith composites

Exploration of outer space sparked interest in Lunar in-situ construction, which shows great advantages in material availability, low cost, and ease of construction. In this work, different weight ratios of graphene were added to regolith powder to investigate their effects on the sintering process and the mechanical properties of the regolith powder. After mixing and compacting, the graphene-regolith composite was sintered under high temperatures to obtain the consolidated solids. The phase distribution and formation of the sintered solids were characterized using energy-dispersive X-ray and X-ray diffraction spectroscopy. The mechanical properties of the modified regolith were studied by the nano and micro-indentation experiments. The results revealed that the dispersion of graphene could not only construct a typical continuous 3D hybrid network structure but also improve the mechanical property of the sintered composite.

Purifying Adversarial Images Using Adversarial Autoencoder With Conditional Normalizing Flows

We present a <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">target-agnostic</i> adversarial autoencoder with conditional normalizing flows specifically designed to, given any <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">unlabeled</i> image dataset, purify adversarial samples into clean images, i.e., remove adversarial noise from the images while preserving their visual quality. In our model interpretation, samples are processed by manifold projection in which the encoder brings the sample back into a posterior data distribution in latent space so that the sample is less likely to be irregular to the learned representation of any target classifier. Normalizing flows conditioned on top of our hybrid network structure and walk-back training are used to deal with common drawbacks of generative model and autoencoder-based approaches: not only the trade-off between compression loss and over-fitting on training data but also the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">structural model dependency</i> on dataset classes and labels. Experiments demonstrated that our proposed model is preferable to existing target-agnostic adversarial defense methods particularly for large and unlabeled image datasets.