The generation of high-quality,realistic face generation has emerged as a key field of research in computer vision.This paper proposes a robust approach that combines a Super-Resolution Generative Adversarial Network(...The generation of high-quality,realistic face generation has emerged as a key field of research in computer vision.This paper proposes a robust approach that combines a Super-Resolution Generative Adversarial Network(SRGAN)with a Pyramid Attention Module(PAM)to enhance the quality of deep face generation.The SRGAN framework is designed to improve the resolution of generated images,addressing common challenges such as blurriness and a lack of intricate details.The Pyramid Attention Module further complements the process by focusing on multi-scale feature extraction,enabling the network to capture finer details and complex facial features more effectively.The proposed method was trained and evaluated over 100 epochs on the CelebA dataset,demonstrating consistent improvements in image quality and a marked decrease in generator and discriminator losses,reflecting the model’s capacity to learn and synthesize high-quality images effectively,given adequate computational resources.Experimental outcome demonstrates that the SRGAN model with PAM module has outperformed,yielding an aggregate discriminator loss of 0.055 for real,0.043 for fake,and a generator loss of 10.58 after training for 100 epochs.The model has yielded an structural similarity index measure of 0.923,that has outperformed the other models that are considered in the current study for analysis.展开更多
Reconfigurable Intelligent Surface(RIS)is regarded as a cutting-edge technology for the development of future wireless communication networks with improved frequency efficiency and reduced energy consumption.This pape...Reconfigurable Intelligent Surface(RIS)is regarded as a cutting-edge technology for the development of future wireless communication networks with improved frequency efficiency and reduced energy consumption.This paper proposes an architecture by combining RIS with Generalized Spatial Modulation(GSM)and then presents a Multi-Residual Deep Neural Network(MR-DNN)scheme,where the active antennas and their transmitted constellation symbols are detected by sub-DNNs in the detection block.Simulation results demonstrate that the proposed MR-DNN detection algorithm performs considerably better than the traditional Zero-Forcing(ZF)and the Minimum Mean Squared Error(MMSE)detection algorithms in terms of Bit Error Rate(BER).Moreover,the MR-DNN detection algorithm has less time complexity than the traditional detection algorithms.展开更多
Modulation recognition has been long investigated in the literature,however,the performance could be severely degraded in multipath fading channels especially for high-order Quadrature Amplitude Modulation(QAM)signals...Modulation recognition has been long investigated in the literature,however,the performance could be severely degraded in multipath fading channels especially for high-order Quadrature Amplitude Modulation(QAM)signals.This could be a critical problem in the broadband maritime wireless communications,where various propagation paths with large differences in the time of arrival are very likely to exist.Specifically,multiple paths may stem from the direct path,the reflection paths from the rough sea surface,and the refraction paths from the atmospheric duct,respectively.To address this issue,we propose a novel blind equalization-aided deep learning(DL)approach to recognize QAM signals in the presence of multipath propagation.The proposed approach consists of two modules:A blind equalization module and a subsequent DL network which employs the structure of ResNet.With predefined searching step-sizes for the blind equalization algorithm,which are designed according to the set of modulation formats of interest,the DL network is trained and tested over various multipath channel parameter settings.It is shown that as compared to the conventional DL approaches without equalization,the proposed method can achieve an improvement in the recognition accuracy up to 30%in severe multipath scenarios,especially in the high SNR regime.Moreover,it efficiently reduces the number of training data that is required.展开更多
Aim: To diagnose COVID-19 more efficiently and more correctly, this study proposed a novel attention network forCOVID-19 (ANC). Methods: Two datasets were used in this study. An 18-way data augmentation was proposed t...Aim: To diagnose COVID-19 more efficiently and more correctly, this study proposed a novel attention network forCOVID-19 (ANC). Methods: Two datasets were used in this study. An 18-way data augmentation was proposed toavoid overfitting. Then, convolutional block attention module (CBAM) was integrated to our model, the structureof which is fine-tuned. Finally, Grad-CAM was used to provide an explainable diagnosis. Results: The accuracyof our ANC methods on two datasets are 96.32% ± 1.06%, and 96.00% ± 1.03%, respectively. Conclusions: Thisproposed ANC method is superior to 9 state-of-the-art approaches.展开更多
Automatic modulation classification(AMC)aims at identifying the modulation of the received signals,which is a significant approach to identifying the target in military and civil applications.In this paper,a novel dat...Automatic modulation classification(AMC)aims at identifying the modulation of the received signals,which is a significant approach to identifying the target in military and civil applications.In this paper,a novel data-driven framework named convolutional and transformer-based deep neural network(CTDNN)is proposed to improve the classification performance.CTDNN can be divided into four modules,i.e.,convolutional neural network(CNN)backbone,transition module,transformer module,and final classifier.In the CNN backbone,a wide and deep convolution structure is designed,which consists of 1×15 convolution kernels and intensive cross-layer connections instead of traditional 1×3 kernels and sequential connections.In the transition module,a 1×1 convolution layer is utilized to compress the channels of the previous multi-scale CNN features.In the transformer module,three self-attention layers are designed for extracting global features and generating the classification vector.In the classifier,the final decision is made based on the maximum a posterior probability.Extensive simulations are conducted,and the result shows that our proposed CTDNN can achieve superior classification performance than traditional deep models.展开更多
With the development of communication systems, modulation methods are becoming more and more diverse. Among them, quadrature spatial modulation(QSM) is considered as one method with less capacity and high efficiency. ...With the development of communication systems, modulation methods are becoming more and more diverse. Among them, quadrature spatial modulation(QSM) is considered as one method with less capacity and high efficiency. In QSM, the traditional signal detection methods sometimes are unable to meet the actual requirement of low complexity of the system. Therefore, this paper proposes a signal detection scheme for QSM systems using deep learning to solve the complexity problem. Results from the simulations show that the bit error rate performance of the proposed deep learning-based detector is better than that of the zero-forcing(ZF) and minimum mean square error(MMSE) detectors, and similar to the maximum likelihood(ML) detector. Moreover, the proposed method requires less processing time than ZF, MMSE,and ML.展开更多
Improving the detection accuracy of rail internal defects and the generalization ability of detection models are not only the main problems in the field of defect detection but also the key to ensuring the safe operat...Improving the detection accuracy of rail internal defects and the generalization ability of detection models are not only the main problems in the field of defect detection but also the key to ensuring the safe operation of high-speed trains.For this reason,a rail internal defect detection method based on an enhanced network structure and module design using ultrasonic images is proposed in this paper.First,a data augmentation method was used to extend the existing image dataset to obtain appropriate image samples.Second,an enhanced network structure was designed to make full use of the high-level and low-level feature information in the image,which improved the accuracy of defect detection.Subsequently,to optimize the detection performance of the proposed model,the Mish activation function was used to design the block module of the feature extraction network.Finally,the pro-posed rail defect detection model was trained.The experimental results showed that the precision rate and F1score of the proposed method were as high as 98%,while the model’s recall rate reached 99%.Specifically,good detec-tion results were achieved for different types of defects,which provides a reference for the engineering application of internal defect detection.Experimental results verified the effectiveness of the proposed method.展开更多
Automatic modulation classification(AMC) technology is one of the cutting-edge technologies in cognitive radio communications. AMC based on deep learning has recently attracted much attention due to its superior perfo...Automatic modulation classification(AMC) technology is one of the cutting-edge technologies in cognitive radio communications. AMC based on deep learning has recently attracted much attention due to its superior performances in classification accuracy and robustness. In this paper, we propose a novel, high resolution and multi-scale feature fusion convolutional neural network model with a squeeze-excitation block, referred to as HRSENet,to classify different kinds of modulation signals.The proposed model establishes a parallel computing mechanism of multi-resolution feature maps through the multi-layer convolution operation, which effectively reduces the information loss caused by downsampling convolution. Moreover, through dense skipconnecting at the same resolution and up-sampling or down-sampling connection at different resolutions, the low resolution representation of the deep feature maps and the high resolution representation of the shallow feature maps are simultaneously extracted and fully integrated, which is benificial to mine signal multilevel features. Finally, the feature squeeze and excitation module embedded in the decoder is used to adjust the response weights between channels, further improving classification accuracy of proposed model.The proposed HRSENet significantly outperforms existing methods in terms of classification accuracy on the public dataset “Over the Air” in signal-to-noise(SNR) ranging from-2dB to 20dB. The classification accuracy in the proposed model achieves 85.36% and97.30% at 4dB and 10dB, respectively, with the improvement by 9.71% and 5.82% compared to LWNet.Furthermore, the model also has a moderate computation complexity compared with several state-of-the-art methods.展开更多
In this paper,a powerful model-driven deep learning framework is exploited to overcome the challenge of multi-domain signal detection in spacedomain index modulation(SDIM)based multiple input multiple output(MIMO)syst...In this paper,a powerful model-driven deep learning framework is exploited to overcome the challenge of multi-domain signal detection in spacedomain index modulation(SDIM)based multiple input multiple output(MIMO)systems.Specifically,we use orthogonal approximate message passing(OAMP)technique to develop OAMPNet,which is a novel signal recovery mechanism in the field of compressed sensing that effectively uses the sparse property from the training SDIM samples.For OAMPNet,the prior probability of the transmit signal has a significant impact on the obtainable performance.For this reason,in our design,we first derive the prior probability of transmitting signals on each antenna for SDIMMIMO systems,which is different from the conventional massive MIMO systems.Then,for massive MIMO scenarios,we propose two novel algorithms to avoid pre-storing all active antenna combinations,thus considerably improving the memory efficiency and reducing the related overhead.Our simulation results show that the proposed framework outperforms the conventional optimization-driven based detection algorithms and has strong robustness under different antenna scales.展开更多
In view of low recognition rate of complex radar intra-pulse modulation signal type by traditional methods under low signal-to-noise ratio(SNR),the paper proposes an automatic recog-nition method of complex radar intr...In view of low recognition rate of complex radar intra-pulse modulation signal type by traditional methods under low signal-to-noise ratio(SNR),the paper proposes an automatic recog-nition method of complex radar intra-pulse modulation signal type based on deep residual network.The basic principle of the recognition method is to obtain the transformation relationship between the time and frequency of complex radar intra-pulse modulation signal through short-time Fourier transform(STFT),and then design an appropriate deep residual network to extract the features of the time-frequency map and complete a variety of complex intra-pulse modulation signal type recognition.In addition,in order to improve the generalization ability of the proposed method,label smoothing and L2 regularization are introduced.The simulation results show that the proposed method has a recognition accuracy of more than 95%for complex radar intra-pulse modulation sig-nal types under low SNR(2 dB).展开更多
In recent times,pattern recognition of communication modulation signals has gained significant attention in several application areas such as military,civilian field,etc.It becomes essential to design a safe and robus...In recent times,pattern recognition of communication modulation signals has gained significant attention in several application areas such as military,civilian field,etc.It becomes essential to design a safe and robust feature extraction(FE)approach to efficiently identify the various signal modulation types in a complex platform.Several works have derived new techniques to extract the feature parameters namely instant features,fractal features,and so on.In addition,machine learning(ML)and deep learning(DL)approaches can be commonly employed for modulation signal classification.In this view,this paper designs pattern recognition of communication signal modulation using fractal features with deep neural networks(CSM-FFDNN).The goal of the CSM-FFDNN model is to classify the different types of digitally modulated signals.The proposed CSM-FFDNN model involves two major processes namely FE and classification.The proposed model uses Sevcik Fractal Dimension(SFD)technique to extract the fractal features from the digital modulated signals.Besides,the extracted features are fed into the DNN model for modulation signal classification.To improve the classification performance of the DNN model,a barnacles mating optimizer(BMO)is used for the hyperparameter tuning of the DNN model in such a way that the DNN performance can be raised.A wide range of simulations takes place to highlight the enhanced performance of the CSM-FFDNN model.The experimental outcomes pointed out the superior recognition rate of the CSM-FFDNN model over the recent state of art methods interms of different evaluation parameters.展开更多
Deep learning techniques are revolutionizing the developmentof medical image segmentation.With the advancement of Transformer models,especially ViT and Swin-Transformer,which enhances the remote-dependent modeling cap...Deep learning techniques are revolutionizing the developmentof medical image segmentation.With the advancement of Transformer models,especially ViT and Swin-Transformer,which enhances the remote-dependent modeling capability of the model through the self-attention mechanism,better segmentation performance can be achieve.Moreover,the high computational cost of Transformer has motivated researchers to explore more efficient models,such as the Mamba model based on state-space modeling(SSM),and for the field of medical segmentation,reducing the number of model parameters is also necessary.In this study,a novel asymmetric model called LA-UMamba was proposed,which integrates visual Mamba module to efficiently capture complex visual features and remote dependencies.The classical design of U-Net was adopted in the upsampling phase to help reduce the number of references and recover more details.To mitigate the information loss problem,an auxiliary U-Net downsampling layer was designed to focus on sizing without extracting features,thus enhancing the protection of input information while maintaining the efficiency of the model.The experiments were conducted on the ACDC MRI cardiac segmentation dataset,and the results showed that the proposed LA-UMamba achieves proved performance compared to the baseline model in several evaluation metrics,such as IoU,Accuracy,Precision,HD and ASD,which improved that the model is successful in optimizing the detail processing and reducing the complexity of the model,providing a new perspective for further optimization of medical image segmentation techniques.展开更多
Medical image classification has played an important role in the medical field, and the related method based on deep learning has become an important and powerful technique in medical image classification. In this art...Medical image classification has played an important role in the medical field, and the related method based on deep learning has become an important and powerful technique in medical image classification. In this article, we propose a simplified inception module based Hadamard attention (SI + HA) mechanism for medical image classification. Specifically, we propose a new attention mechanism: Hadamard attention mechanism. It improves the accuracy of medical image classification without greatly increasing the complexity of the model. Meanwhile, we adopt a simplified inception module to improve the utilization of parameters. We use two medical image datasets to prove the superiority of our proposed method. In the BreakHis dataset, the AUCs of our method can reach 98.74%, 98.38%, 98.61% and 97.67% under the magnification factors of 40×, 100×, 200× and 400×, respectively. The accuracies can reach 95.67%, 94.17%, 94.53% and 94.12% under the magnification factors of 40×, 100×, 200× and 400×, respectively. In the KIMIA Path 960 dataset, the AUCs and accuracy of our method can reach 99.91% and 99.03%. It is superior to the currently popular methods and can significantly improve the effectiveness of medical image classification.展开更多
Although modulation classification based on deep neural network can achieve high Modulation Classification(MC)accuracies,catastrophic forgetting will occur when the neural network model continues to learn new tasks.In...Although modulation classification based on deep neural network can achieve high Modulation Classification(MC)accuracies,catastrophic forgetting will occur when the neural network model continues to learn new tasks.In this paper,we simulate the dynamic wireless communication environment and focus on breaking the learning paradigm of isolated automatic MC.We innovate a research algorithm for continuous automatic MC.Firstly,a memory for storing representative old task modulation signals is built,which is employed to limit the gradient update direction of new tasks in the continuous learning stage to ensure that the loss of old tasks is also in a downward trend.Secondly,in order to better simulate the dynamic wireless communication environment,we employ the mini-batch gradient algorithm which is more suitable for continuous learning.Finally,the signal in the memory can be replayed to further strengthen the characteristics of the old task signal in the model.Simulation results verify the effectiveness of the method.展开更多
Deep forgery detection technologies are crucial for image and video recognition tasks,with their performance heavily reliant on the features extracted from both real and fake images.However,most existing methods prima...Deep forgery detection technologies are crucial for image and video recognition tasks,with their performance heavily reliant on the features extracted from both real and fake images.However,most existing methods primarily focus on spatial domain features,which limits their accuracy.To address this limitation,we propose an adaptive dual-domain feature representation method for enhanced deep forgery detection.Specifically,an adaptive region dynamic convolution module is established to efficiently extract facial features from the spatial domain.Then,we introduce an adaptive frequency dynamic filter to capture effective frequency domain features.By fusing both spatial and frequency domain features,our approach significantly improves the accuracy of classifying real and fake facial images.Finally,experimental results on three real-world datasets validate the effectiveness of our dual-domain feature representation method,which substantially improves classification precision.展开更多
The utilization of millimeter-wave frequencies and cognitive radio(CR)are promising ways to increase the spectral efficiency of wireless communication systems.However,conventional CR spectrum sensing techniques entail...The utilization of millimeter-wave frequencies and cognitive radio(CR)are promising ways to increase the spectral efficiency of wireless communication systems.However,conventional CR spectrum sensing techniques entail sampling the received signal at a Nyquist rate,and they are not viable for wideband signals due to their high cost.This paper expounds on how sub-Nyquist sampling in conjunction with deep learning can be leveraged to remove this limitation.To this end,we propose a multi-task learning(MTL)framework using convolutional neural networks for the joint inference of the underlying narrowband signal number,their modulation scheme,and their location in a wideband spectrum.We demonstrate the effectiveness of the proposed framework for real-world millimeter-wave wideband signals collected by physical devices,exhibiting a 91.7% accuracy in the joint inference task when considering up to two narrowband signals over a wideband spectrum.Ultimately,the proposed data-driven approach enables on-the-fly wideband spectrum sensing,combining accuracy,and computational efficiency,which are indispensable for CR and opportunistic networking.展开更多
针对复杂电子对抗场景中的非理想信道环境,该文提出了一种基于深度学习的异常检测(anomaly detection,AD)方法。首先,分析了利用时频同相/正交(in-phase/quadrature,I/Q)采样数据进行AD的可行性;然后,设计了深度学习网络架构,并提出基...针对复杂电子对抗场景中的非理想信道环境,该文提出了一种基于深度学习的异常检测(anomaly detection,AD)方法。首先,分析了利用时频同相/正交(in-phase/quadrature,I/Q)采样数据进行AD的可行性;然后,设计了深度学习网络架构,并提出基于深度支持向量描述(deep support vector data description,Deep SVDD)和调制识别的AD方法。仿真及实验结果表明:相比于经典的单分类检测算法,该方法检测性能和实时性明显提升,且在非理想信道环境下表现鲁棒。该方法已在某型号项目原理样机上得到验证,具有很高应用价值。展开更多
基金supported by the National Research Foundation of Korea(NRF)grant funded by the Korea government(*MSIT)(No.2018R1A5A7059549).
文摘The generation of high-quality,realistic face generation has emerged as a key field of research in computer vision.This paper proposes a robust approach that combines a Super-Resolution Generative Adversarial Network(SRGAN)with a Pyramid Attention Module(PAM)to enhance the quality of deep face generation.The SRGAN framework is designed to improve the resolution of generated images,addressing common challenges such as blurriness and a lack of intricate details.The Pyramid Attention Module further complements the process by focusing on multi-scale feature extraction,enabling the network to capture finer details and complex facial features more effectively.The proposed method was trained and evaluated over 100 epochs on the CelebA dataset,demonstrating consistent improvements in image quality and a marked decrease in generator and discriminator losses,reflecting the model’s capacity to learn and synthesize high-quality images effectively,given adequate computational resources.Experimental outcome demonstrates that the SRGAN model with PAM module has outperformed,yielding an aggregate discriminator loss of 0.055 for real,0.043 for fake,and a generator loss of 10.58 after training for 100 epochs.The model has yielded an structural similarity index measure of 0.923,that has outperformed the other models that are considered in the current study for analysis.
基金supported in part by the Shenzhen Basic Research Program under Grant JCYJ20220531103008018,20231120142345001 and 20231127144045001the Guangdong Basic Research Program under Grant 2024ZDZX1016the Natural Science Foundation of China under Grant U20A20156.
文摘Reconfigurable Intelligent Surface(RIS)is regarded as a cutting-edge technology for the development of future wireless communication networks with improved frequency efficiency and reduced energy consumption.This paper proposes an architecture by combining RIS with Generalized Spatial Modulation(GSM)and then presents a Multi-Residual Deep Neural Network(MR-DNN)scheme,where the active antennas and their transmitted constellation symbols are detected by sub-DNNs in the detection block.Simulation results demonstrate that the proposed MR-DNN detection algorithm performs considerably better than the traditional Zero-Forcing(ZF)and the Minimum Mean Squared Error(MMSE)detection algorithms in terms of Bit Error Rate(BER).Moreover,the MR-DNN detection algorithm has less time complexity than the traditional detection algorithms.
基金the National Natural Science Foundation of China under Grant 61771264,61801114,61501264,61771286the Nantong University-Nantong Joint Research Center for Intelligent Information Technology under Grant No.KFKT2017B01,KFKT2017A04the Natural Science Foundation of Jiangsu Province under Grant BK20170688.
文摘Modulation recognition has been long investigated in the literature,however,the performance could be severely degraded in multipath fading channels especially for high-order Quadrature Amplitude Modulation(QAM)signals.This could be a critical problem in the broadband maritime wireless communications,where various propagation paths with large differences in the time of arrival are very likely to exist.Specifically,multiple paths may stem from the direct path,the reflection paths from the rough sea surface,and the refraction paths from the atmospheric duct,respectively.To address this issue,we propose a novel blind equalization-aided deep learning(DL)approach to recognize QAM signals in the presence of multipath propagation.The proposed approach consists of two modules:A blind equalization module and a subsequent DL network which employs the structure of ResNet.With predefined searching step-sizes for the blind equalization algorithm,which are designed according to the set of modulation formats of interest,the DL network is trained and tested over various multipath channel parameter settings.It is shown that as compared to the conventional DL approaches without equalization,the proposed method can achieve an improvement in the recognition accuracy up to 30%in severe multipath scenarios,especially in the high SNR regime.Moreover,it efficiently reduces the number of training data that is required.
基金This paper is partially supported by Open Fund for Jiangsu Key Laboratory of Advanced Manufacturing Technology(HGAMTL-1703)Guangxi Key Laboratory of Trusted Software(kx201901)+5 种基金Fundamental Research Funds for the Central Universities(CDLS-2020-03)Key Laboratory of Child Development and Learning Science(Southeast University),Ministry of EducationRoyal Society International Exchanges Cost Share Award,UK(RP202G0230)Medical Research Council Confidence in Concept Award,UK(MC_PC_17171)Hope Foundation for Cancer Research,UK(RM60G0680)British Heart Foundation Accelerator Award,UK.
文摘Aim: To diagnose COVID-19 more efficiently and more correctly, this study proposed a novel attention network forCOVID-19 (ANC). Methods: Two datasets were used in this study. An 18-way data augmentation was proposed toavoid overfitting. Then, convolutional block attention module (CBAM) was integrated to our model, the structureof which is fine-tuned. Finally, Grad-CAM was used to provide an explainable diagnosis. Results: The accuracyof our ANC methods on two datasets are 96.32% ± 1.06%, and 96.00% ± 1.03%, respectively. Conclusions: Thisproposed ANC method is superior to 9 state-of-the-art approaches.
基金supported in part by the National Natural Science Foundation of China under Grant(62171045,62201090)in part by the National Key Research and Development Program of China under Grants(2020YFB1807602,2019YFB1804404).
文摘Automatic modulation classification(AMC)aims at identifying the modulation of the received signals,which is a significant approach to identifying the target in military and civil applications.In this paper,a novel data-driven framework named convolutional and transformer-based deep neural network(CTDNN)is proposed to improve the classification performance.CTDNN can be divided into four modules,i.e.,convolutional neural network(CNN)backbone,transition module,transformer module,and final classifier.In the CNN backbone,a wide and deep convolution structure is designed,which consists of 1×15 convolution kernels and intensive cross-layer connections instead of traditional 1×3 kernels and sequential connections.In the transition module,a 1×1 convolution layer is utilized to compress the channels of the previous multi-scale CNN features.In the transformer module,three self-attention layers are designed for extracting global features and generating the classification vector.In the classifier,the final decision is made based on the maximum a posterior probability.Extensive simulations are conducted,and the result shows that our proposed CTDNN can achieve superior classification performance than traditional deep models.
基金supported in part by The Science and Technology Development Fund, Macao SAR, China (0108/2020/A3)in part by The Science and Technology Development Fund, Macao SAR, China (0005/2021/ITP)the Deanship of Scientific Research at Taif University for funding this work。
文摘With the development of communication systems, modulation methods are becoming more and more diverse. Among them, quadrature spatial modulation(QSM) is considered as one method with less capacity and high efficiency. In QSM, the traditional signal detection methods sometimes are unable to meet the actual requirement of low complexity of the system. Therefore, this paper proposes a signal detection scheme for QSM systems using deep learning to solve the complexity problem. Results from the simulations show that the bit error rate performance of the proposed deep learning-based detector is better than that of the zero-forcing(ZF) and minimum mean square error(MMSE) detectors, and similar to the maximum likelihood(ML) detector. Moreover, the proposed method requires less processing time than ZF, MMSE,and ML.
基金Supported by National Natural Science Foundation of China(Grant No.61573233)Guangdong Provincial Natural Science Foundation of China(Grant No.2021A1515010661)Guangdong Provincial Special Projects in Key Fields of Colleges and Universities of China(Grant No.2020ZDZX2005).
文摘Improving the detection accuracy of rail internal defects and the generalization ability of detection models are not only the main problems in the field of defect detection but also the key to ensuring the safe operation of high-speed trains.For this reason,a rail internal defect detection method based on an enhanced network structure and module design using ultrasonic images is proposed in this paper.First,a data augmentation method was used to extend the existing image dataset to obtain appropriate image samples.Second,an enhanced network structure was designed to make full use of the high-level and low-level feature information in the image,which improved the accuracy of defect detection.Subsequently,to optimize the detection performance of the proposed model,the Mish activation function was used to design the block module of the feature extraction network.Finally,the pro-posed rail defect detection model was trained.The experimental results showed that the precision rate and F1score of the proposed method were as high as 98%,while the model’s recall rate reached 99%.Specifically,good detec-tion results were achieved for different types of defects,which provides a reference for the engineering application of internal defect detection.Experimental results verified the effectiveness of the proposed method.
基金supported by the Beijing Natural Science Foundation (L202003)National Natural Science Foundation of China (No. 31700479)。
文摘Automatic modulation classification(AMC) technology is one of the cutting-edge technologies in cognitive radio communications. AMC based on deep learning has recently attracted much attention due to its superior performances in classification accuracy and robustness. In this paper, we propose a novel, high resolution and multi-scale feature fusion convolutional neural network model with a squeeze-excitation block, referred to as HRSENet,to classify different kinds of modulation signals.The proposed model establishes a parallel computing mechanism of multi-resolution feature maps through the multi-layer convolution operation, which effectively reduces the information loss caused by downsampling convolution. Moreover, through dense skipconnecting at the same resolution and up-sampling or down-sampling connection at different resolutions, the low resolution representation of the deep feature maps and the high resolution representation of the shallow feature maps are simultaneously extracted and fully integrated, which is benificial to mine signal multilevel features. Finally, the feature squeeze and excitation module embedded in the decoder is used to adjust the response weights between channels, further improving classification accuracy of proposed model.The proposed HRSENet significantly outperforms existing methods in terms of classification accuracy on the public dataset “Over the Air” in signal-to-noise(SNR) ranging from-2dB to 20dB. The classification accuracy in the proposed model achieves 85.36% and97.30% at 4dB and 10dB, respectively, with the improvement by 9.71% and 5.82% compared to LWNet.Furthermore, the model also has a moderate computation complexity compared with several state-of-the-art methods.
基金supported by the National Natural Science Foundation of China under Grant U19B2014the Sichuan Science and Technology Program under Grant 2023NSFSC0457the Fundamental Research Funds for the Central Universities under Grant 2242022k60006.
文摘In this paper,a powerful model-driven deep learning framework is exploited to overcome the challenge of multi-domain signal detection in spacedomain index modulation(SDIM)based multiple input multiple output(MIMO)systems.Specifically,we use orthogonal approximate message passing(OAMP)technique to develop OAMPNet,which is a novel signal recovery mechanism in the field of compressed sensing that effectively uses the sparse property from the training SDIM samples.For OAMPNet,the prior probability of the transmit signal has a significant impact on the obtainable performance.For this reason,in our design,we first derive the prior probability of transmitting signals on each antenna for SDIMMIMO systems,which is different from the conventional massive MIMO systems.Then,for massive MIMO scenarios,we propose two novel algorithms to avoid pre-storing all active antenna combinations,thus considerably improving the memory efficiency and reducing the related overhead.Our simulation results show that the proposed framework outperforms the conventional optimization-driven based detection algorithms and has strong robustness under different antenna scales.
文摘In view of low recognition rate of complex radar intra-pulse modulation signal type by traditional methods under low signal-to-noise ratio(SNR),the paper proposes an automatic recog-nition method of complex radar intra-pulse modulation signal type based on deep residual network.The basic principle of the recognition method is to obtain the transformation relationship between the time and frequency of complex radar intra-pulse modulation signal through short-time Fourier transform(STFT),and then design an appropriate deep residual network to extract the features of the time-frequency map and complete a variety of complex intra-pulse modulation signal type recognition.In addition,in order to improve the generalization ability of the proposed method,label smoothing and L2 regularization are introduced.The simulation results show that the proposed method has a recognition accuracy of more than 95%for complex radar intra-pulse modulation sig-nal types under low SNR(2 dB).
基金supported by the National Research Foundation of Korea(NRF)grant funded by the Korea government(MSIT)(No.2021R1F1A1063319).
文摘In recent times,pattern recognition of communication modulation signals has gained significant attention in several application areas such as military,civilian field,etc.It becomes essential to design a safe and robust feature extraction(FE)approach to efficiently identify the various signal modulation types in a complex platform.Several works have derived new techniques to extract the feature parameters namely instant features,fractal features,and so on.In addition,machine learning(ML)and deep learning(DL)approaches can be commonly employed for modulation signal classification.In this view,this paper designs pattern recognition of communication signal modulation using fractal features with deep neural networks(CSM-FFDNN).The goal of the CSM-FFDNN model is to classify the different types of digitally modulated signals.The proposed CSM-FFDNN model involves two major processes namely FE and classification.The proposed model uses Sevcik Fractal Dimension(SFD)technique to extract the fractal features from the digital modulated signals.Besides,the extracted features are fed into the DNN model for modulation signal classification.To improve the classification performance of the DNN model,a barnacles mating optimizer(BMO)is used for the hyperparameter tuning of the DNN model in such a way that the DNN performance can be raised.A wide range of simulations takes place to highlight the enhanced performance of the CSM-FFDNN model.The experimental outcomes pointed out the superior recognition rate of the CSM-FFDNN model over the recent state of art methods interms of different evaluation parameters.
文摘Deep learning techniques are revolutionizing the developmentof medical image segmentation.With the advancement of Transformer models,especially ViT and Swin-Transformer,which enhances the remote-dependent modeling capability of the model through the self-attention mechanism,better segmentation performance can be achieve.Moreover,the high computational cost of Transformer has motivated researchers to explore more efficient models,such as the Mamba model based on state-space modeling(SSM),and for the field of medical segmentation,reducing the number of model parameters is also necessary.In this study,a novel asymmetric model called LA-UMamba was proposed,which integrates visual Mamba module to efficiently capture complex visual features and remote dependencies.The classical design of U-Net was adopted in the upsampling phase to help reduce the number of references and recover more details.To mitigate the information loss problem,an auxiliary U-Net downsampling layer was designed to focus on sizing without extracting features,thus enhancing the protection of input information while maintaining the efficiency of the model.The experiments were conducted on the ACDC MRI cardiac segmentation dataset,and the results showed that the proposed LA-UMamba achieves proved performance compared to the baseline model in several evaluation metrics,such as IoU,Accuracy,Precision,HD and ASD,which improved that the model is successful in optimizing the detail processing and reducing the complexity of the model,providing a new perspective for further optimization of medical image segmentation techniques.
文摘Medical image classification has played an important role in the medical field, and the related method based on deep learning has become an important and powerful technique in medical image classification. In this article, we propose a simplified inception module based Hadamard attention (SI + HA) mechanism for medical image classification. Specifically, we propose a new attention mechanism: Hadamard attention mechanism. It improves the accuracy of medical image classification without greatly increasing the complexity of the model. Meanwhile, we adopt a simplified inception module to improve the utilization of parameters. We use two medical image datasets to prove the superiority of our proposed method. In the BreakHis dataset, the AUCs of our method can reach 98.74%, 98.38%, 98.61% and 97.67% under the magnification factors of 40×, 100×, 200× and 400×, respectively. The accuracies can reach 95.67%, 94.17%, 94.53% and 94.12% under the magnification factors of 40×, 100×, 200× and 400×, respectively. In the KIMIA Path 960 dataset, the AUCs and accuracy of our method can reach 99.91% and 99.03%. It is superior to the currently popular methods and can significantly improve the effectiveness of medical image classification.
文摘Although modulation classification based on deep neural network can achieve high Modulation Classification(MC)accuracies,catastrophic forgetting will occur when the neural network model continues to learn new tasks.In this paper,we simulate the dynamic wireless communication environment and focus on breaking the learning paradigm of isolated automatic MC.We innovate a research algorithm for continuous automatic MC.Firstly,a memory for storing representative old task modulation signals is built,which is employed to limit the gradient update direction of new tasks in the continuous learning stage to ensure that the loss of old tasks is also in a downward trend.Secondly,in order to better simulate the dynamic wireless communication environment,we employ the mini-batch gradient algorithm which is more suitable for continuous learning.Finally,the signal in the memory can be replayed to further strengthen the characteristics of the old task signal in the model.Simulation results verify the effectiveness of the method.
基金supported in part by the National Natural Science Foundation of China under No.12401679the Nature Science Foundation of the Jiangsu Higher Education Institutions of China under No.23KJB520006the Haizhou Bay Talent Innovation Program of Jiangsu Ocean University under No.PD2024026。
文摘Deep forgery detection technologies are crucial for image and video recognition tasks,with their performance heavily reliant on the features extracted from both real and fake images.However,most existing methods primarily focus on spatial domain features,which limits their accuracy.To address this limitation,we propose an adaptive dual-domain feature representation method for enhanced deep forgery detection.Specifically,an adaptive region dynamic convolution module is established to efficiently extract facial features from the spatial domain.Then,we introduce an adaptive frequency dynamic filter to capture effective frequency domain features.By fusing both spatial and frequency domain features,our approach significantly improves the accuracy of classifying real and fake facial images.Finally,experimental results on three real-world datasets validate the effectiveness of our dual-domain feature representation method,which substantially improves classification precision.
文摘The utilization of millimeter-wave frequencies and cognitive radio(CR)are promising ways to increase the spectral efficiency of wireless communication systems.However,conventional CR spectrum sensing techniques entail sampling the received signal at a Nyquist rate,and they are not viable for wideband signals due to their high cost.This paper expounds on how sub-Nyquist sampling in conjunction with deep learning can be leveraged to remove this limitation.To this end,we propose a multi-task learning(MTL)framework using convolutional neural networks for the joint inference of the underlying narrowband signal number,their modulation scheme,and their location in a wideband spectrum.We demonstrate the effectiveness of the proposed framework for real-world millimeter-wave wideband signals collected by physical devices,exhibiting a 91.7% accuracy in the joint inference task when considering up to two narrowband signals over a wideband spectrum.Ultimately,the proposed data-driven approach enables on-the-fly wideband spectrum sensing,combining accuracy,and computational efficiency,which are indispensable for CR and opportunistic networking.
文摘针对复杂电子对抗场景中的非理想信道环境,该文提出了一种基于深度学习的异常检测(anomaly detection,AD)方法。首先,分析了利用时频同相/正交(in-phase/quadrature,I/Q)采样数据进行AD的可行性;然后,设计了深度学习网络架构,并提出基于深度支持向量描述(deep support vector data description,Deep SVDD)和调制识别的AD方法。仿真及实验结果表明:相比于经典的单分类检测算法,该方法检测性能和实时性明显提升,且在非理想信道环境下表现鲁棒。该方法已在某型号项目原理样机上得到验证,具有很高应用价值。
