Skin diseases affect millions worldwide.Early detection is key to preventing disfigurement,lifelong disability,or death.Dermoscopic images acquired in primary-care settings show high intra-class visual similarity and ...Skin diseases affect millions worldwide.Early detection is key to preventing disfigurement,lifelong disability,or death.Dermoscopic images acquired in primary-care settings show high intra-class visual similarity and severe class imbalance,and occasional imaging artifacts can create ambiguity for state-of-the-art convolutional neural networks(CNNs).We frame skin lesion recognition as graph-based reasoning and,to ensure fair evaluation and avoid data leakage,adopt a strict lesion-level partitioning strategy.Each image is first over-segmented using SLIC(Simple Linear Iterative Clustering)to produce perceptually homogeneous superpixels.These superpixels form the nodes of a region-adjacency graph whose edges encode spatial continuity.Node attributes are 1280-dimensional embeddings extracted with a lightweight yet expressive EfficientNet-B0 backbone,providing strong representational power at modest computational cost.The resulting graphs are processed by a five-layer Graph Attention Network(GAT)that learns to weight inter-node relationships dynamically and aggregates multi-hop context before classifying lesions into seven classes with a log-softmax output.Extensive experiments on the DermaMNIST benchmark show the proposed pipeline achieves 88.35%accuracy and 98.04%AUC,outperforming contemporary CNNs,AutoML approaches,and alternative graph neural networks.An ablation study indicates EfficientNet-B0 produces superior node descriptors compared with ResNet-18 and DenseNet,and that roughly five GAT layers strike a good balance between being too shallow and over-deep while avoiding oversmoothing.The method requires no data augmentation or external metadata,making it a drop-in upgrade for clinical computer-aided diagnosis systems.展开更多
With the increasing severity of network security threats,Network Intrusion Detection(NID)has become a key technology to ensure network security.To address the problem of low detection rate of traditional intrusion det...With the increasing severity of network security threats,Network Intrusion Detection(NID)has become a key technology to ensure network security.To address the problem of low detection rate of traditional intrusion detection models,this paper proposes a Dual-Attention model for NID,which combines Convolutional Neural Network(CNN)and Bidirectional Long Short-Term Memory(BiLSTM)to design two modules:the FocusConV and the TempoNet module.The FocusConV module,which automatically adjusts and weights CNN extracted local features,focuses on local features that are more important for intrusion detection.The TempoNet module focuses on global information,identifies more important features in time steps or sequences,and filters and weights the information globally to further improve the accuracy and robustness of NID.Meanwhile,in order to solve the class imbalance problem in the dataset,the EQL v2 method is used to compute the class weights of each class and to use them in the loss computation,which optimizes the performance of the model on the class imbalance problem.Extensive experiments were conducted on the NSL-KDD,UNSW-NB15,and CIC-DDos2019 datasets,achieving average accuracy rates of 99.66%,87.47%,and 99.39%,respectively,demonstrating excellent detection accuracy and robustness.The model also improves the detection performance of minority classes in the datasets.On the UNSW-NB15 dataset,the detection rates for Analysis,Exploits,and Shellcode attacks increased by 7%,7%,and 10%,respectively,demonstrating the Dual-Attention CNN-BiLSTM model’s excellent performance in NID.展开更多
Real-time prediction and precise control of sinter quality are pivotal for energy saving,cost reduction,quality improvement and efficiency enhancement in the ironmaking process.To advance,the accuracy and comprehensiv...Real-time prediction and precise control of sinter quality are pivotal for energy saving,cost reduction,quality improvement and efficiency enhancement in the ironmaking process.To advance,the accuracy and comprehensiveness of sinter quality prediction,an intelligent flare monitoring system for sintering machine tails that combines hybrid neural networks integrating convolutional neural network with long short-term memory(CNN-LSTM)networks was proposed.The system utilized a high-temperature thermal imager for image acquisition at the sintering machine tail and employed a zone-triggered method to accurately capture dynamic feature images under challenging conditions of high-temperature,high dust,and occlusion.The feature images were then segmented through a triple-iteration multi-thresholding approach based on the maximum between-class variance method to minimize detail loss during the segmentation process.Leveraging the advantages of CNN and LSTM networks in capturing temporal and spatial information,a comprehensive model for sinter quality prediction was constructed,with inputs including the proportion of combustion layer,porosity rate,temperature distribution,and image features obtained from the convolutional neural network,and outputs comprising quality indicators such as underburning index,uniformity index,and FeO content of the sinter.The accuracy is notably increased,achieving a 95.8%hit rate within an error margin of±1.0.After the system is applied,the average qualified rate of FeO content increases from 87.24%to 89.99%,representing an improvement of 2.75%.The average monthly solid fuel consumption is reduced from 49.75 to 46.44 kg/t,leading to a 6.65%reduction and underscoring significant energy saving and cost reduction effects.展开更多
Skin cancer is the most prevalent cancer globally,primarily due to extensive exposure to Ultraviolet(UV)radiation.Early identification of skin cancer enhances the likelihood of effective treatment,as delays may lead t...Skin cancer is the most prevalent cancer globally,primarily due to extensive exposure to Ultraviolet(UV)radiation.Early identification of skin cancer enhances the likelihood of effective treatment,as delays may lead to severe tumor advancement.This study proposes a novel hybrid deep learning strategy to address the complex issue of skin cancer diagnosis,with an architecture that integrates a Vision Transformer,a bespoke convolutional neural network(CNN),and an Xception module.They were evaluated using two benchmark datasets,HAM10000 and Skin Cancer ISIC.On the HAM10000,the model achieves a precision of 95.46%,an accuracy of 96.74%,a recall of 96.27%,specificity of 96.00%and an F1-Score of 95.86%.It obtains an accuracy of 93.19%,a precision of 93.25%,a recall of 92.80%,a specificity of 92.89%and an F1-Score of 93.19%on the Skin Cancer ISIC dataset.The findings demonstrate that the model that was proposed is robust and trustworthy when it comes to the classification of skin lesions.In addition,the utilization of Explainable AI techniques,such as Grad-CAM visualizations,assists in highlighting the most significant lesion areas that have an impact on the decisions that are made by the model.展开更多
Micro-expressions(ME)recognition is a complex task that requires advanced techniques to extract informative features fromfacial expressions.Numerous deep neural networks(DNNs)with convolutional structures have been pr...Micro-expressions(ME)recognition is a complex task that requires advanced techniques to extract informative features fromfacial expressions.Numerous deep neural networks(DNNs)with convolutional structures have been proposed.However,unlike DNNs,shallow convolutional neural networks often outperform deeper models in mitigating overfitting,particularly with small datasets.Still,many of these methods rely on a single feature for recognition,resulting in an insufficient ability to extract highly effective features.To address this limitation,in this paper,an Improved Dual-stream Shallow Convolutional Neural Network based on an Extreme Gradient Boosting Algorithm(IDSSCNN-XgBoost)is introduced for ME Recognition.The proposed method utilizes a dual-stream architecture where motion vectors(temporal features)are extracted using Optical Flow TV-L1 and amplify subtle changes(spatial features)via EulerianVideoMagnification(EVM).These features are processed by IDSSCNN,with an attention mechanism applied to refine the extracted effective features.The outputs are then fused,concatenated,and classified using the XgBoost algorithm.This comprehensive approach significantly improves recognition accuracy by leveraging the strengths of both temporal and spatial information,supported by the robust classification power of XgBoost.The proposed method is evaluated on three publicly available ME databases named Chinese Academy of Sciences Micro-expression Database(CASMEII),Spontaneous Micro-Expression Database(SMICHS),and Spontaneous Actions and Micro-Movements(SAMM).Experimental results indicate that the proposed model can achieve outstanding results compared to recent models.The accuracy results are 79.01%,69.22%,and 68.99%on CASMEII,SMIC-HS,and SAMM,and the F1-score are 75.47%,68.91%,and 63.84%,respectively.The proposed method has the advantage of operational efficiency and less computational time.展开更多
In the field of image processing,the analysis of Synthetic Aperture Radar(SAR)images is crucial due to its broad range of applications.However,SAR images are often affected by coherent speckle noise,which significantl...In the field of image processing,the analysis of Synthetic Aperture Radar(SAR)images is crucial due to its broad range of applications.However,SAR images are often affected by coherent speckle noise,which significantly degrades image quality.Traditional denoising methods,typically based on filter techniques,often face challenges related to inefficiency and limited adaptability.To address these limitations,this study proposes a novel SAR image denoising algorithm based on an enhanced residual network architecture,with the objective of enhancing the utility of SAR imagery in complex electromagnetic environments.The proposed algorithm integrates residual network modules,which directly process the noisy input images to generate denoised outputs.This approach not only reduces computational complexity but also mitigates the difficulties associated with model training.By combining the Transformer module with the residual block,the algorithm enhances the network's ability to extract global features,offering superior feature extraction capabilities compared to CNN-based residual modules.Additionally,the algorithm employs the adaptive activation function Meta-ACON,which dynamically adjusts the activation patterns of neurons,thereby improving the network's feature extraction efficiency.The effectiveness of the proposed denoising method is empirically validated using real SAR images from the RSOD dataset.The proposed algorithm exhibits remarkable performance in terms of EPI,SSIM,and ENL,while achieving a substantial enhancement in PSNR when compared to traditional and deep learning-based algorithms.The PSNR performance is enhanced by over twofold.Moreover,the evaluation of the MSTAR SAR dataset substantiates the algorithm's robustness and applicability in SAR denoising tasks,with a PSNR of 25.2021 being attained.These findings underscore the efficacy of the proposed algorithm in mitigating speckle noise while preserving critical features in SAR imagery,thereby enhancing its quality and usability in practical scenarios.展开更多
Forecasting energy demand is essential for optimizing energy generation and effectively predicting power system needs.Recently,many researchers have developed various models on tabular datasets to enhance the effectiv...Forecasting energy demand is essential for optimizing energy generation and effectively predicting power system needs.Recently,many researchers have developed various models on tabular datasets to enhance the effectiveness of demand prediction,including neural networks,machine learning,deep learning,and advanced architectures such as CNN and LSTM.However,research on the CNN models has struggled to provide reliable outcomes due to insufficient dataset sizes,repeated investigations,and inappropriate baseline selection.To address these challenges,we propose a Tabular data-based Lightweight Convolutional Neural Network(TLCNN)model for predicting energy demand.It frames the problem as a regression task that effectively captures complex data trends for accurate forecasting.The BanE-16 dataset is preprocessed using normalization techniques for categorical and numerical data before training the model.The proposed approach dynamically selects relevant features through a two-dimensional convolutional structure that improves adaptability.The model’s performance is evaluated using MSE,MAE,and Accuracy metrics.Experimental results show that TLCNN achieves a 10.89%lower MSE than traditional ML algorithms,demonstrating superior predictive capability.Additionally,TLCNN’s lightweight structure enhances generalization while reducing computational costs,making it suitable for real-world energy forecasting tasks.This study contributes to energy informatics by introducing an optimized deep-learning framework that improves demand prediction by ensuring robustness and adaptability for tabular data.展开更多
Skin cancer remains a significant global health challenge,and early detection is crucial to improving patient outcomes.This study presents a novel deep learning framework that combines Convolutional Neural Networks(CN...Skin cancer remains a significant global health challenge,and early detection is crucial to improving patient outcomes.This study presents a novel deep learning framework that combines Convolutional Neural Networks(CNNs),Transformers,and Gated Recurrent Units(GRUs)for robust skin cancer classification.To address data set imbalance,we employ StyleGAN3-based synthetic data augmentation alongside traditional techniques.The hybrid architecture effectively captures both local and global dependencies in dermoscopic images,while the GRU component models sequential patterns.Evaluated on the HAM10000 dataset,the proposed model achieves an accuracy of 90.61%,outperforming baseline architectures such as VGG16 and ResNet.Our system also demonstrates superior precision(91.11%),recall(95.28%),and AUC(0.97),highlighting its potential as a reliable diagnostic tool for the detection of melanoma.This work advances automated skin cancer diagnosis by addressing critical challenges related to class imbalance and limited generalization in medical imaging.展开更多
Hyperspectral image(HSI)classification is crucial for numerous remote sensing applications.Traditional deep learning methods may miss pixel relationships and context,leading to inefficiencies.This paper introduces the...Hyperspectral image(HSI)classification is crucial for numerous remote sensing applications.Traditional deep learning methods may miss pixel relationships and context,leading to inefficiencies.This paper introduces the spectral band graph convolutional and attention-enhanced CNN joint network(SGCCN),a novel approach that harnesses the power of spectral band graph convolutions for capturing long-range relationships,utilizes local perception of attention-enhanced multi-level convolutions for local spatial feature and employs a dynamic attention mechanism to enhance feature extraction.The SGCCN integrates spectral and spatial features through a self-attention fusion network,significantly improving classification accuracy and efficiency.The proposed method outperforms existing techniques,demonstrating its effectiveness in handling the challenges associated with HSI data.展开更多
Deep convolutional neural networks(CNNs)have demonstrated remarkable performance in video super-resolution(VSR).However,the ability of most existing methods to recover fine details in complex scenes is often hindered ...Deep convolutional neural networks(CNNs)have demonstrated remarkable performance in video super-resolution(VSR).However,the ability of most existing methods to recover fine details in complex scenes is often hindered by the loss of shallow texture information during feature extraction.To address this limitation,we propose a 3D Convolutional Enhanced Residual Video Super-Resolution Network(3D-ERVSNet).This network employs a forward and backward bidirectional propagation module(FBBPM)that aligns features across frames using explicit optical flow through lightweight SPyNet.By incorporating an enhanced residual structure(ERS)with skip connections,shallow and deep features are effectively integrated,enhancing texture restoration capabilities.Furthermore,3D convolution module(3DCM)is applied after the backward propagation module to implicitly capture spatio-temporal dependencies.The architecture synergizes these components where FBBPM extracts aligned features,ERS fuses hierarchical representations,and 3DCM refines temporal coherence.Finally,a deep feature aggregation module(DFAM)fuses the processed features,and a pixel-upsampling module(PUM)reconstructs the high-resolution(HR)video frames.Comprehensive evaluations on REDS,Vid4,UDM10,and Vim4 benchmarks demonstrate well performance including 30.95 dB PSNR/0.8822 SSIM on REDS and 32.78 dB/0.8987 on Vim4.3D-ERVSNet achieves significant gains over baselines while maintaining high efficiency with only 6.3M parameters and 77ms/frame runtime(i.e.,20×faster than RBPN).The network’s effectiveness stems from its task-specific asymmetric design that balances explicit alignment and implicit fusion.展开更多
Fluorescence microscopy is indispensable in life science research,yet denoising remains challenging due to varied biological samples and imaging conditions.We introduce a wavelet-enhanced transformer based on DnCNN th...Fluorescence microscopy is indispensable in life science research,yet denoising remains challenging due to varied biological samples and imaging conditions.We introduce a wavelet-enhanced transformer based on DnCNN that fuses wavelet preprocessing with a dual-branch transformer-convolutional neural network(CNN)architecture.Wavelet decomposition separates highand low-frequency components for targeted noise reduction;the CNN branch restores local details,whereas the transformer branch captures global context;and an adaptive loss balances quantitative fidelity with perceptual quality.On the fluorescence microscopy denoising benchmark,our method surpasses leading CNNand transformer-based approaches,improving peak signal-to-noise ratio by 2.34%and 0.88%and structural similarity index measure by 0.53%and 1.07%,respectively.This framework offers enhanced generalization and practical gains for fluorescence image denoising.展开更多
Scleral vessels on the surface of the human eye can provide valuable information about potential diseases or dysfunctions of specific organs,and vessel segmentation is a key step in characterizing the scleral vessels....Scleral vessels on the surface of the human eye can provide valuable information about potential diseases or dysfunctions of specific organs,and vessel segmentation is a key step in characterizing the scleral vessels.However,accurate segmentation of blood vessels in the scleral images is a challenging task due to the intricate texture,tenuous structure,and erratic network of the scleral vessels.In this work,we propose a CNN-Transformer hybrid network named SVSNet for automatic scleral vessel segmentation.Following the typical U-shape encoder-decoder architecture,the SVSNet integrates a Sobel edge detection module to provide edge prior and further combines the Atrous Spatial Pyramid Pooling module to enhance its ability to extract vessels of various sizes.At the end of the encoding path,a vision Transformer module is incorporated to capture the global context and improve the continuity of the vessel network.To validate the effectiveness of the proposed SVSNet,comparative experiments are conducted on two public scleral image datasets,and the results show that the SVSNet outperforms other state-of-the-art models.Further experiments on three public retinal image datasets demonstrate that the SVSNet can be easily applied to other vessel datasets with good generalization capability.展开更多
基金funded by the Deanship of Graduate Studies and Scientific Research at Jouf University under grant No.(DGSSR-2025-02-01296).
文摘Skin diseases affect millions worldwide.Early detection is key to preventing disfigurement,lifelong disability,or death.Dermoscopic images acquired in primary-care settings show high intra-class visual similarity and severe class imbalance,and occasional imaging artifacts can create ambiguity for state-of-the-art convolutional neural networks(CNNs).We frame skin lesion recognition as graph-based reasoning and,to ensure fair evaluation and avoid data leakage,adopt a strict lesion-level partitioning strategy.Each image is first over-segmented using SLIC(Simple Linear Iterative Clustering)to produce perceptually homogeneous superpixels.These superpixels form the nodes of a region-adjacency graph whose edges encode spatial continuity.Node attributes are 1280-dimensional embeddings extracted with a lightweight yet expressive EfficientNet-B0 backbone,providing strong representational power at modest computational cost.The resulting graphs are processed by a five-layer Graph Attention Network(GAT)that learns to weight inter-node relationships dynamically and aggregates multi-hop context before classifying lesions into seven classes with a log-softmax output.Extensive experiments on the DermaMNIST benchmark show the proposed pipeline achieves 88.35%accuracy and 98.04%AUC,outperforming contemporary CNNs,AutoML approaches,and alternative graph neural networks.An ablation study indicates EfficientNet-B0 produces superior node descriptors compared with ResNet-18 and DenseNet,and that roughly five GAT layers strike a good balance between being too shallow and over-deep while avoiding oversmoothing.The method requires no data augmentation or external metadata,making it a drop-in upgrade for clinical computer-aided diagnosis systems.
基金supported by the High-Level Talent Foundation of Jinling Institute of Technology(grant number.JIT-B-202413).
文摘With the increasing severity of network security threats,Network Intrusion Detection(NID)has become a key technology to ensure network security.To address the problem of low detection rate of traditional intrusion detection models,this paper proposes a Dual-Attention model for NID,which combines Convolutional Neural Network(CNN)and Bidirectional Long Short-Term Memory(BiLSTM)to design two modules:the FocusConV and the TempoNet module.The FocusConV module,which automatically adjusts and weights CNN extracted local features,focuses on local features that are more important for intrusion detection.The TempoNet module focuses on global information,identifies more important features in time steps or sequences,and filters and weights the information globally to further improve the accuracy and robustness of NID.Meanwhile,in order to solve the class imbalance problem in the dataset,the EQL v2 method is used to compute the class weights of each class and to use them in the loss computation,which optimizes the performance of the model on the class imbalance problem.Extensive experiments were conducted on the NSL-KDD,UNSW-NB15,and CIC-DDos2019 datasets,achieving average accuracy rates of 99.66%,87.47%,and 99.39%,respectively,demonstrating excellent detection accuracy and robustness.The model also improves the detection performance of minority classes in the datasets.On the UNSW-NB15 dataset,the detection rates for Analysis,Exploits,and Shellcode attacks increased by 7%,7%,and 10%,respectively,demonstrating the Dual-Attention CNN-BiLSTM model’s excellent performance in NID.
基金founded by the Open Project Program of Anhui Province Key Laboratory of Metallurgical Engineering and Resources Recycling(Anhui University of Technology)(No.SKF21-06)Research Fund for Young Teachers of Anhui University of Technology in 2020(No.QZ202001).
文摘Real-time prediction and precise control of sinter quality are pivotal for energy saving,cost reduction,quality improvement and efficiency enhancement in the ironmaking process.To advance,the accuracy and comprehensiveness of sinter quality prediction,an intelligent flare monitoring system for sintering machine tails that combines hybrid neural networks integrating convolutional neural network with long short-term memory(CNN-LSTM)networks was proposed.The system utilized a high-temperature thermal imager for image acquisition at the sintering machine tail and employed a zone-triggered method to accurately capture dynamic feature images under challenging conditions of high-temperature,high dust,and occlusion.The feature images were then segmented through a triple-iteration multi-thresholding approach based on the maximum between-class variance method to minimize detail loss during the segmentation process.Leveraging the advantages of CNN and LSTM networks in capturing temporal and spatial information,a comprehensive model for sinter quality prediction was constructed,with inputs including the proportion of combustion layer,porosity rate,temperature distribution,and image features obtained from the convolutional neural network,and outputs comprising quality indicators such as underburning index,uniformity index,and FeO content of the sinter.The accuracy is notably increased,achieving a 95.8%hit rate within an error margin of±1.0.After the system is applied,the average qualified rate of FeO content increases from 87.24%to 89.99%,representing an improvement of 2.75%.The average monthly solid fuel consumption is reduced from 49.75 to 46.44 kg/t,leading to a 6.65%reduction and underscoring significant energy saving and cost reduction effects.
文摘Skin cancer is the most prevalent cancer globally,primarily due to extensive exposure to Ultraviolet(UV)radiation.Early identification of skin cancer enhances the likelihood of effective treatment,as delays may lead to severe tumor advancement.This study proposes a novel hybrid deep learning strategy to address the complex issue of skin cancer diagnosis,with an architecture that integrates a Vision Transformer,a bespoke convolutional neural network(CNN),and an Xception module.They were evaluated using two benchmark datasets,HAM10000 and Skin Cancer ISIC.On the HAM10000,the model achieves a precision of 95.46%,an accuracy of 96.74%,a recall of 96.27%,specificity of 96.00%and an F1-Score of 95.86%.It obtains an accuracy of 93.19%,a precision of 93.25%,a recall of 92.80%,a specificity of 92.89%and an F1-Score of 93.19%on the Skin Cancer ISIC dataset.The findings demonstrate that the model that was proposed is robust and trustworthy when it comes to the classification of skin lesions.In addition,the utilization of Explainable AI techniques,such as Grad-CAM visualizations,assists in highlighting the most significant lesion areas that have an impact on the decisions that are made by the model.
基金supported by the Key Research and Development Program of Jiangsu Province under Grant BE2022059-3,CTBC Bank through the Industry-Academia Cooperation Project,as well as by the Ministry of Science and Technology of Taiwan through Grants MOST-108-2218-E-002-055,MOST-109-2223-E-009-002-MY3,MOST-109-2218-E-009-025,and MOST431109-2218-E-002-015.
文摘Micro-expressions(ME)recognition is a complex task that requires advanced techniques to extract informative features fromfacial expressions.Numerous deep neural networks(DNNs)with convolutional structures have been proposed.However,unlike DNNs,shallow convolutional neural networks often outperform deeper models in mitigating overfitting,particularly with small datasets.Still,many of these methods rely on a single feature for recognition,resulting in an insufficient ability to extract highly effective features.To address this limitation,in this paper,an Improved Dual-stream Shallow Convolutional Neural Network based on an Extreme Gradient Boosting Algorithm(IDSSCNN-XgBoost)is introduced for ME Recognition.The proposed method utilizes a dual-stream architecture where motion vectors(temporal features)are extracted using Optical Flow TV-L1 and amplify subtle changes(spatial features)via EulerianVideoMagnification(EVM).These features are processed by IDSSCNN,with an attention mechanism applied to refine the extracted effective features.The outputs are then fused,concatenated,and classified using the XgBoost algorithm.This comprehensive approach significantly improves recognition accuracy by leveraging the strengths of both temporal and spatial information,supported by the robust classification power of XgBoost.The proposed method is evaluated on three publicly available ME databases named Chinese Academy of Sciences Micro-expression Database(CASMEII),Spontaneous Micro-Expression Database(SMICHS),and Spontaneous Actions and Micro-Movements(SAMM).Experimental results indicate that the proposed model can achieve outstanding results compared to recent models.The accuracy results are 79.01%,69.22%,and 68.99%on CASMEII,SMIC-HS,and SAMM,and the F1-score are 75.47%,68.91%,and 63.84%,respectively.The proposed method has the advantage of operational efficiency and less computational time.
文摘In the field of image processing,the analysis of Synthetic Aperture Radar(SAR)images is crucial due to its broad range of applications.However,SAR images are often affected by coherent speckle noise,which significantly degrades image quality.Traditional denoising methods,typically based on filter techniques,often face challenges related to inefficiency and limited adaptability.To address these limitations,this study proposes a novel SAR image denoising algorithm based on an enhanced residual network architecture,with the objective of enhancing the utility of SAR imagery in complex electromagnetic environments.The proposed algorithm integrates residual network modules,which directly process the noisy input images to generate denoised outputs.This approach not only reduces computational complexity but also mitigates the difficulties associated with model training.By combining the Transformer module with the residual block,the algorithm enhances the network's ability to extract global features,offering superior feature extraction capabilities compared to CNN-based residual modules.Additionally,the algorithm employs the adaptive activation function Meta-ACON,which dynamically adjusts the activation patterns of neurons,thereby improving the network's feature extraction efficiency.The effectiveness of the proposed denoising method is empirically validated using real SAR images from the RSOD dataset.The proposed algorithm exhibits remarkable performance in terms of EPI,SSIM,and ENL,while achieving a substantial enhancement in PSNR when compared to traditional and deep learning-based algorithms.The PSNR performance is enhanced by over twofold.Moreover,the evaluation of the MSTAR SAR dataset substantiates the algorithm's robustness and applicability in SAR denoising tasks,with a PSNR of 25.2021 being attained.These findings underscore the efficacy of the proposed algorithm in mitigating speckle noise while preserving critical features in SAR imagery,thereby enhancing its quality and usability in practical scenarios.
文摘Forecasting energy demand is essential for optimizing energy generation and effectively predicting power system needs.Recently,many researchers have developed various models on tabular datasets to enhance the effectiveness of demand prediction,including neural networks,machine learning,deep learning,and advanced architectures such as CNN and LSTM.However,research on the CNN models has struggled to provide reliable outcomes due to insufficient dataset sizes,repeated investigations,and inappropriate baseline selection.To address these challenges,we propose a Tabular data-based Lightweight Convolutional Neural Network(TLCNN)model for predicting energy demand.It frames the problem as a regression task that effectively captures complex data trends for accurate forecasting.The BanE-16 dataset is preprocessed using normalization techniques for categorical and numerical data before training the model.The proposed approach dynamically selects relevant features through a two-dimensional convolutional structure that improves adaptability.The model’s performance is evaluated using MSE,MAE,and Accuracy metrics.Experimental results show that TLCNN achieves a 10.89%lower MSE than traditional ML algorithms,demonstrating superior predictive capability.Additionally,TLCNN’s lightweight structure enhances generalization while reducing computational costs,making it suitable for real-world energy forecasting tasks.This study contributes to energy informatics by introducing an optimized deep-learning framework that improves demand prediction by ensuring robustness and adaptability for tabular data.
文摘Skin cancer remains a significant global health challenge,and early detection is crucial to improving patient outcomes.This study presents a novel deep learning framework that combines Convolutional Neural Networks(CNNs),Transformers,and Gated Recurrent Units(GRUs)for robust skin cancer classification.To address data set imbalance,we employ StyleGAN3-based synthetic data augmentation alongside traditional techniques.The hybrid architecture effectively captures both local and global dependencies in dermoscopic images,while the GRU component models sequential patterns.Evaluated on the HAM10000 dataset,the proposed model achieves an accuracy of 90.61%,outperforming baseline architectures such as VGG16 and ResNet.Our system also demonstrates superior precision(91.11%),recall(95.28%),and AUC(0.97),highlighting its potential as a reliable diagnostic tool for the detection of melanoma.This work advances automated skin cancer diagnosis by addressing critical challenges related to class imbalance and limited generalization in medical imaging.
基金supported in part by the National Natural Science Foundations of China(No.61801214)the Postgraduate Research Practice Innovation Program of NUAA(No.xcxjh20231504)。
文摘Hyperspectral image(HSI)classification is crucial for numerous remote sensing applications.Traditional deep learning methods may miss pixel relationships and context,leading to inefficiencies.This paper introduces the spectral band graph convolutional and attention-enhanced CNN joint network(SGCCN),a novel approach that harnesses the power of spectral band graph convolutions for capturing long-range relationships,utilizes local perception of attention-enhanced multi-level convolutions for local spatial feature and employs a dynamic attention mechanism to enhance feature extraction.The SGCCN integrates spectral and spatial features through a self-attention fusion network,significantly improving classification accuracy and efficiency.The proposed method outperforms existing techniques,demonstrating its effectiveness in handling the challenges associated with HSI data.
基金supported in part by the Basic and Applied Basic Research Foundation of Guangdong Province[2025A1515011566]in part by the State Key Laboratory for Novel Software Technology,Nanjing University[KFKT2024B08]+1 种基金in part by Leading Talents in Gusu Innovation and Entrepreneurship[ZXL2023170]in part by the Basic Research Programs of Taicang 2024,[TC2024JC32].
文摘Deep convolutional neural networks(CNNs)have demonstrated remarkable performance in video super-resolution(VSR).However,the ability of most existing methods to recover fine details in complex scenes is often hindered by the loss of shallow texture information during feature extraction.To address this limitation,we propose a 3D Convolutional Enhanced Residual Video Super-Resolution Network(3D-ERVSNet).This network employs a forward and backward bidirectional propagation module(FBBPM)that aligns features across frames using explicit optical flow through lightweight SPyNet.By incorporating an enhanced residual structure(ERS)with skip connections,shallow and deep features are effectively integrated,enhancing texture restoration capabilities.Furthermore,3D convolution module(3DCM)is applied after the backward propagation module to implicitly capture spatio-temporal dependencies.The architecture synergizes these components where FBBPM extracts aligned features,ERS fuses hierarchical representations,and 3DCM refines temporal coherence.Finally,a deep feature aggregation module(DFAM)fuses the processed features,and a pixel-upsampling module(PUM)reconstructs the high-resolution(HR)video frames.Comprehensive evaluations on REDS,Vid4,UDM10,and Vim4 benchmarks demonstrate well performance including 30.95 dB PSNR/0.8822 SSIM on REDS and 32.78 dB/0.8987 on Vim4.3D-ERVSNet achieves significant gains over baselines while maintaining high efficiency with only 6.3M parameters and 77ms/frame runtime(i.e.,20×faster than RBPN).The network’s effectiveness stems from its task-specific asymmetric design that balances explicit alignment and implicit fusion.
基金supported by the National Natural Science Foundation of China(Grant No.62275210)the National Leading Talent Program,the National Young Talent Program,the Key Research and Development Program of Shaanxi(Grant No.2024SF2-GJHX-25)+5 种基金the Scientific Research Program Funded by the Education Department of Shaanxi Provincial Government(Grant No.24JS016)the Xidian University Specially Funded Project for Interdisciplinary Exploration(Grant No.TZJHF202523)the Fundamental Research Funds for Central Universities(Grant No.YJSJ25014)the Guangdong Provincial General Colleges and Universities Young Innovative Talents Research Project(Grant No.2024KQNCX172)the Shenzhen Science and Technology Program(Grant No.GJHZ20210705141805015)the Key Research Areas Support Science and Technology Project of Shenzhen Institute of Information Technology(Grant No.SZIIT2024KJ056).
文摘Fluorescence microscopy is indispensable in life science research,yet denoising remains challenging due to varied biological samples and imaging conditions.We introduce a wavelet-enhanced transformer based on DnCNN that fuses wavelet preprocessing with a dual-branch transformer-convolutional neural network(CNN)architecture.Wavelet decomposition separates highand low-frequency components for targeted noise reduction;the CNN branch restores local details,whereas the transformer branch captures global context;and an adaptive loss balances quantitative fidelity with perceptual quality.On the fluorescence microscopy denoising benchmark,our method surpasses leading CNNand transformer-based approaches,improving peak signal-to-noise ratio by 2.34%and 0.88%and structural similarity index measure by 0.53%and 1.07%,respectively.This framework offers enhanced generalization and practical gains for fluorescence image denoising.
基金supported by the National Key Research and Development Program of China(2022YFC3502301 and 2022YFC3502300)the National Natural Science Foundation of China(52475546)+1 种基金the R&D Program of Beijing Municipal Education Commission(KM202311232021)the Young Backbone Teacher Support Plan of Beijing Information Science&Technology University(YBT202410).
文摘Scleral vessels on the surface of the human eye can provide valuable information about potential diseases or dysfunctions of specific organs,and vessel segmentation is a key step in characterizing the scleral vessels.However,accurate segmentation of blood vessels in the scleral images is a challenging task due to the intricate texture,tenuous structure,and erratic network of the scleral vessels.In this work,we propose a CNN-Transformer hybrid network named SVSNet for automatic scleral vessel segmentation.Following the typical U-shape encoder-decoder architecture,the SVSNet integrates a Sobel edge detection module to provide edge prior and further combines the Atrous Spatial Pyramid Pooling module to enhance its ability to extract vessels of various sizes.At the end of the encoding path,a vision Transformer module is incorporated to capture the global context and improve the continuity of the vessel network.To validate the effectiveness of the proposed SVSNet,comparative experiments are conducted on two public scleral image datasets,and the results show that the SVSNet outperforms other state-of-the-art models.Further experiments on three public retinal image datasets demonstrate that the SVSNet can be easily applied to other vessel datasets with good generalization capability.
