Discriminative region localization and efficient feature encoding are crucial for fine-grained object recognition.However,existing data augmentation methods struggle to accurately locate discriminative regions in comp...Discriminative region localization and efficient feature encoding are crucial for fine-grained object recognition.However,existing data augmentation methods struggle to accurately locate discriminative regions in complex backgrounds,small target objects,and limited training data,leading to poor recognition.Fine-grained images exhibit“small inter-class differences,”and while second-order feature encoding enhances discrimination,it often requires dual Convolutional Neural Networks(CNN),increasing training time and complexity.This study proposes a model integrating discriminative region localization and efficient second-order feature encoding.By ranking feature map channels via a fully connected layer,it selects high-importance channels to generate an enhanced map,accurately locating discriminative regions.Cropping and erasing augmentations further refine recognition.To improve efficiency,a novel second-order feature encoding module generates an attention map from the fourth convolutional group of Residual Network 50 layers(ResNet-50)and multiplies it with features from the fifth group,producing second-order features while reducing dimensionality and training time.Experiments on Caltech-University of California,San Diego Birds-200-2011(CUB-200-2011),Stanford Car,and Fine-Grained Visual Classification of Aircraft(FGVC Aircraft)datasets show state-of-the-art accuracy of 88.9%,94.7%,and 93.3%,respectively.展开更多
Deep learning(DL)methods like multilayer perceptrons(MLPs)and convolutional neural networks(CNNs)have been applied to predict the complex traits in animal and plant breeding.However,improving the genomic prediction ac...Deep learning(DL)methods like multilayer perceptrons(MLPs)and convolutional neural networks(CNNs)have been applied to predict the complex traits in animal and plant breeding.However,improving the genomic prediction accuracy still presents signifcant challenges.In this study,we applied CNNs to predict swine traits using previously published data.Specifcally,we extensively evaluated the CNN model's performance by employing various sets of single nucleotide polymorphisms(SNPs)and concluded that the CNN model achieved optimal performance when utilizing SNP sets comprising 1,000 SNPs.Furthermore,we adopted a novel approach using the one-hot encoding method that transforms the 16 different genotypes into sets of eight binary variables.This innovative encoding method signifcantly enhanced the CNN's prediction accuracy for swine traits,outperforming the traditional one-hot encoding techniques.Our fndings suggest that the expanded one-hot encoding method can improve the accuracy of DL methods in the genomic prediction of swine agricultural economic traits.This discovery has significant implications for swine breeding programs,where genomic prediction is pivotal in improving breeding strategies.Furthermore,future research endeavors can explore additional enhancements to DL methods by incorporating advanced data pre-processing techniques.展开更多
With the rapid expansion of social media,analyzing emotions and their causes in texts has gained significant importance.Emotion-cause pair extraction enables the identification of causal relationships between emotions...With the rapid expansion of social media,analyzing emotions and their causes in texts has gained significant importance.Emotion-cause pair extraction enables the identification of causal relationships between emotions and their triggers within a text,facilitating a deeper understanding of expressed sentiments and their underlying reasons.This comprehension is crucial for making informed strategic decisions in various business and societal contexts.However,recent research approaches employing multi-task learning frameworks for modeling often face challenges such as the inability to simultaneouslymodel extracted features and their interactions,or inconsistencies in label prediction between emotion-cause pair extraction and independent assistant tasks like emotion and cause extraction.To address these issues,this study proposes an emotion-cause pair extraction methodology that incorporates joint feature encoding and task alignment mechanisms.The model consists of two primary components:First,joint feature encoding simultaneously generates features for emotion-cause pairs and clauses,enhancing feature interactions between emotion clauses,cause clauses,and emotion-cause pairs.Second,the task alignment technique is applied to reduce the labeling distance between emotion-cause pair extraction and the two assistant tasks,capturing deep semantic information interactions among tasks.The proposed method is evaluated on a Chinese benchmark corpus using 10-fold cross-validation,assessing key performance metrics such as precision,recall,and F1 score.Experimental results demonstrate that the model achieves an F1 score of 76.05%,surpassing the state-of-the-art by 1.03%.The proposed model exhibits significant improvements in emotion-cause pair extraction(ECPE)and cause extraction(CE)compared to existing methods,validating its effectiveness.This research introduces a novel approach based on joint feature encoding and task alignment mechanisms,contributing to advancements in emotion-cause pair extraction.However,the study’s limitation lies in the data sources,potentially restricting the generalizability of the findings.展开更多
Retinal blood vessel segmentation is crucial for diagnosing ocular and cardiovascular diseases.Although the introduction of U-Net in 2015 by Olaf Ronneberger significantly advanced this field,yet issues like limited t...Retinal blood vessel segmentation is crucial for diagnosing ocular and cardiovascular diseases.Although the introduction of U-Net in 2015 by Olaf Ronneberger significantly advanced this field,yet issues like limited training data,imbalance data distribution,and inadequate feature extraction persist,hindering both the segmentation performance and optimal model generalization.Addressing these critical issues,the DEFFA-Unet is proposed featuring an additional encoder to process domain-invariant pre-processed inputs,thereby improving both richer feature encoding and enhanced model generalization.A feature filtering fusion module is developed to ensure the precise feature filtering and robust hybrid feature fusion.In response to the task-specific need for higher precision where false positives are very costly,traditional skip connections are replaced with the attention-guided feature reconstructing fusion module.Additionally,innovative data augmentation and balancing methods are proposed to counter data scarcity and distribution imbalance,further boosting the robustness and generalization of the model.With a comprehensive suite of evaluation metrics,extensive validations on four benchmark datasets(DRIVE,CHASEDB1,STARE,and HRF)and an SLO dataset(IOSTAR),demonstrate the proposed method’s superiority over both baseline and state-of-the-art models.Particularly the proposed method significantly outperforms the compared methods in cross-validation model generalization.展开更多
Quantum communication networks,such as quantum key distribution(QKD)networks,typically employ the measurement-resend mechanism between two users using quantum communication devices based on different quantum encoding ...Quantum communication networks,such as quantum key distribution(QKD)networks,typically employ the measurement-resend mechanism between two users using quantum communication devices based on different quantum encoding types.To achieve direct communication between the devices with different quantum encoding types,in this paper,we propose encoding conversion schemes between the polarization bases(rectilinear,diagonal and circular bases)and the time-bin phase bases(two phase bases and time-bin basis)and design the quantum encoding converters.The theoretical analysis of the encoding conversion schemes is given in detail,and the basis correspondence of encoding conversion and the property of bit flip are revealed.The conversion relationship between polarization bases and time-bin phase bases can be easily selected by controlling a phase shifter.Since no optical switches are used in our scheme,the converter can be operated with high speed.The converters can also be modularized,which may be utilized to realize miniaturization in the future.展开更多
Blockchain,as a distributed ledger,inherently possesses tamper-resistant capabilities,creating a natural channel for covert communication.However,the immutable nature of data storage might introduce challenges to comm...Blockchain,as a distributed ledger,inherently possesses tamper-resistant capabilities,creating a natural channel for covert communication.However,the immutable nature of data storage might introduce challenges to communication security.This study introduces a blockchain-based covert communication model utilizing dynamic Base-K encoding.The proposed encoding scheme utilizes the input address sequence to determine K to encode the secret message and determines the order of transactions based on K,thus ensuring effective concealment of the message.The dynamic encoding parameters enhance flexibility and address issues related to identical transaction amounts for the same secret message.Experimental results demonstrate that the proposed method maintains smooth communication and low susceptibility to tampering,achieving commendable concealment and embedding rates.展开更多
High-Speed Trains (HSTs) have emerged as a mainstream mode of transportation in China, owing to their exceptional safety and efficiency. Ensuring the reliable operation of HSTs is of paramount economic and societal im...High-Speed Trains (HSTs) have emerged as a mainstream mode of transportation in China, owing to their exceptional safety and efficiency. Ensuring the reliable operation of HSTs is of paramount economic and societal importance. As critical rotating mechanical components of the transmission system, bearings make their fault diagnosis a topic of extensive attention. This paper provides a systematic review of image encoding-based bearing fault diagnosis methods tailored to the condition monitoring of HSTs. First, it categorizes the image encoding techniques applied in the field of bearing fault diagnosis. Then, a review of state-of-the-art studies has been presented, encompassing both monomodal image conversion and multimodal image fusion approaches. Finally, it highlights current challenges and proposes future research directions to advance intelligent fault diagnosis in HSTs, aiming to provide a valuable reference for researchers and engineers in the field of intelligent operation and maintenance.展开更多
The Gaussian phase distribution approximation enables analysis of restricted diffusion encoded by general gradient waveforms but fails to account for the diffraction-like features that may occur for simple pore geomet...The Gaussian phase distribution approximation enables analysis of restricted diffusion encoded by general gradient waveforms but fails to account for the diffraction-like features that may occur for simple pore geometries.We investigate the range of validity of the approximation by random walk simulations of restricted diffusion in a cylinder using isotropic diffusion encoding sequences as well as conventional single gradient pulse pairs and oscillating gradient waveforms.The results show that clear deviations from the approximation may be observed at relative signal attenuations below 0.1 for onedimensional sequences with few oscillation periods.Increasing the encoding dimensionality and/or number of oscillations while extending the total duration of the waveform diminishes the non-Gaussian effects while preserving the low apparent diffusivities characteristic of restriction.展开更多
Faulty-feeder detection in neutral point noneffectively grounded distribution networks consistently attracts research attention since it directly affects quality and safety of energy supply.Most modern research on fau...Faulty-feeder detection in neutral point noneffectively grounded distribution networks consistently attracts research attention since it directly affects quality and safety of energy supply.Most modern research on faulty-feeder detection tends to apply more complex digital signal processing techniques and deeper neural networks in order to better extract and learn as many detailed characteristics as possible.However,these approaches may easily result in overfitting and high computational cost,which cannot meet requirements for detection accuracy and efficiency in practical applications.This paper proposes an innovative waveform encoding method and details a simple convolutional neural network(CNN)with one layer of convolution used for identification,which seeks to improve detection accuracy and efficiency simultaneously.First,sparse characteristics of waveforms are utilized to encode into compact vectors,and a waveform-vector matrix is generated.Second,to deduce waveform-vector matrix,a simple CNN with multi-scale filters and one layer of convolution is established.Finally,a methodology for faulty-feeder detection is proposed,and both detection accuracy and efficiency are considerably enhanced.Comparative studies have confirmed clear superiority of the developed method,which outperforms existing approaches in both detection accuracy and efficiency,thus highlighting its significant potential for application.展开更多
Multimodal sentiment analysis aims to understand emotions from text,speech,and video data.However,current methods often overlook the dominant role of text and suffer from feature loss during integration.Given the vary...Multimodal sentiment analysis aims to understand emotions from text,speech,and video data.However,current methods often overlook the dominant role of text and suffer from feature loss during integration.Given the varying importance of each modality across different contexts,a central and pressing challenge in multimodal sentiment analysis lies in maximizing the use of rich intra-modal features while minimizing information loss during the fusion process.In response to these critical limitations,we propose a novel framework that integrates spatial position encoding and fusion embedding modules to address these issues.In our model,text is treated as the core modality,while speech and video features are selectively incorporated through a unique position-aware fusion process.The spatial position encoding strategy preserves the internal structural information of speech and visual modalities,enabling the model to capture localized intra-modal dependencies that are often overlooked.This design enhances the richness and discriminative power of the fused representation,enabling more accurate and context-aware sentiment prediction.Finally,we conduct comprehensive evaluations on two widely recognized standard datasets in the field—CMU-MOSI and CMU-MOSEI to validate the performance of the proposed model.The experimental results demonstrate that our model exhibits good performance and effectiveness for sentiment analysis tasks.展开更多
Highly programmable shape morphing of 4D-printed micro/nanostructures is urgently desired for applications in robotics and intelligent systems.However,due to the lack of autonomous holistic strategies throughout the t...Highly programmable shape morphing of 4D-printed micro/nanostructures is urgently desired for applications in robotics and intelligent systems.However,due to the lack of autonomous holistic strategies throughout the target shape input,optimal material distribution generation,and fabrication program output,4D nanoprinting that permits arbitrary shape morphing remains a challenging task for manual design.In this study,we report an autonomous inverse encoding strategy to decipher the genetic code for material property distributions that can guide the encoded modeling toward arbitrarily pre-programmed 4D shape morphing.By tuning the laser power of each voxel at the nanoscale,the genetic code can be spatially programmed and controllable shape morphing can be realized through the inverse encoding process.Using this strategy,the 4D-printed structures can be designed and accurately shift to the target morphing of arbitrarily hand-drawn lines under stimulation.Furthermore,as a proof-of-concept,a flexible fiber micromanipulator that can approach the target region through pre-programmed shape morphing is autonomously inversely encoded according to the localized spatial environment.This strategy may contribute to the modeling and arbitrary shape morphing of micro/nanostructures fabricated via 4D nanoprinting,leading to cutting-edge applications in microfluidics,micro-robotics,minimally invasive robotic surgery,and tissue engineering.展开更多
Sensitivity encoding(SENSE)is a parallel magnetic resonance imaging(MRI)reconstruction model by utilizing the sensitivity information of receiver coils to achieve image reconstruction.The existing SENSE-based reconstr...Sensitivity encoding(SENSE)is a parallel magnetic resonance imaging(MRI)reconstruction model by utilizing the sensitivity information of receiver coils to achieve image reconstruction.The existing SENSE-based reconstruction algorithms usually used nonadaptive sparsifying transforms,resulting in a limited reconstruction accuracy.Therefore,we proposed a new model for accurate parallel MRI reconstruction by combining the L0 norm regularization term based on the efficient sum of outer products dictionary learning(SOUPDIL)with the SENSE model,called SOUPDIL-SENSE.The SOUPDIL-SENSE model is mainly solved by utilizing the variable splitting and alternating direction method of multipliers techniques.The experimental results on four human datasets show that the proposed algorithm effectively promotes the image sparsity,eliminates the noise and artifacts of the reconstructed images,and improves the reconstruction accuracy.展开更多
基金supported,in part,by the National Nature Science Foundation of China under Grant 62272236,62376128 and 62306139the Natural Science Foundation of Jiangsu Province under Grant BK20201136,BK20191401.
文摘Discriminative region localization and efficient feature encoding are crucial for fine-grained object recognition.However,existing data augmentation methods struggle to accurately locate discriminative regions in complex backgrounds,small target objects,and limited training data,leading to poor recognition.Fine-grained images exhibit“small inter-class differences,”and while second-order feature encoding enhances discrimination,it often requires dual Convolutional Neural Networks(CNN),increasing training time and complexity.This study proposes a model integrating discriminative region localization and efficient second-order feature encoding.By ranking feature map channels via a fully connected layer,it selects high-importance channels to generate an enhanced map,accurately locating discriminative regions.Cropping and erasing augmentations further refine recognition.To improve efficiency,a novel second-order feature encoding module generates an attention map from the fourth convolutional group of Residual Network 50 layers(ResNet-50)and multiplies it with features from the fifth group,producing second-order features while reducing dimensionality and training time.Experiments on Caltech-University of California,San Diego Birds-200-2011(CUB-200-2011),Stanford Car,and Fine-Grained Visual Classification of Aircraft(FGVC Aircraft)datasets show state-of-the-art accuracy of 88.9%,94.7%,and 93.3%,respectively.
基金supported by the National Natural Science Foundation of China(32102513)the National Key Scientific Research Project(2023YFF1001100)+1 种基金the Shenzhen Innovation and Entrepreneurship PlanMajor Special Project of Science and Technology,China(KJZD20230923115003006)the Innovation Project of Chinese Academy of Agricultural Sciences(CAAS-ZDRW202006)。
文摘Deep learning(DL)methods like multilayer perceptrons(MLPs)and convolutional neural networks(CNNs)have been applied to predict the complex traits in animal and plant breeding.However,improving the genomic prediction accuracy still presents signifcant challenges.In this study,we applied CNNs to predict swine traits using previously published data.Specifcally,we extensively evaluated the CNN model's performance by employing various sets of single nucleotide polymorphisms(SNPs)and concluded that the CNN model achieved optimal performance when utilizing SNP sets comprising 1,000 SNPs.Furthermore,we adopted a novel approach using the one-hot encoding method that transforms the 16 different genotypes into sets of eight binary variables.This innovative encoding method signifcantly enhanced the CNN's prediction accuracy for swine traits,outperforming the traditional one-hot encoding techniques.Our fndings suggest that the expanded one-hot encoding method can improve the accuracy of DL methods in the genomic prediction of swine agricultural economic traits.This discovery has significant implications for swine breeding programs,where genomic prediction is pivotal in improving breeding strategies.Furthermore,future research endeavors can explore additional enhancements to DL methods by incorporating advanced data pre-processing techniques.
文摘With the rapid expansion of social media,analyzing emotions and their causes in texts has gained significant importance.Emotion-cause pair extraction enables the identification of causal relationships between emotions and their triggers within a text,facilitating a deeper understanding of expressed sentiments and their underlying reasons.This comprehension is crucial for making informed strategic decisions in various business and societal contexts.However,recent research approaches employing multi-task learning frameworks for modeling often face challenges such as the inability to simultaneouslymodel extracted features and their interactions,or inconsistencies in label prediction between emotion-cause pair extraction and independent assistant tasks like emotion and cause extraction.To address these issues,this study proposes an emotion-cause pair extraction methodology that incorporates joint feature encoding and task alignment mechanisms.The model consists of two primary components:First,joint feature encoding simultaneously generates features for emotion-cause pairs and clauses,enhancing feature interactions between emotion clauses,cause clauses,and emotion-cause pairs.Second,the task alignment technique is applied to reduce the labeling distance between emotion-cause pair extraction and the two assistant tasks,capturing deep semantic information interactions among tasks.The proposed method is evaluated on a Chinese benchmark corpus using 10-fold cross-validation,assessing key performance metrics such as precision,recall,and F1 score.Experimental results demonstrate that the model achieves an F1 score of 76.05%,surpassing the state-of-the-art by 1.03%.The proposed model exhibits significant improvements in emotion-cause pair extraction(ECPE)and cause extraction(CE)compared to existing methods,validating its effectiveness.This research introduces a novel approach based on joint feature encoding and task alignment mechanisms,contributing to advancements in emotion-cause pair extraction.However,the study’s limitation lies in the data sources,potentially restricting the generalizability of the findings.
文摘Retinal blood vessel segmentation is crucial for diagnosing ocular and cardiovascular diseases.Although the introduction of U-Net in 2015 by Olaf Ronneberger significantly advanced this field,yet issues like limited training data,imbalance data distribution,and inadequate feature extraction persist,hindering both the segmentation performance and optimal model generalization.Addressing these critical issues,the DEFFA-Unet is proposed featuring an additional encoder to process domain-invariant pre-processed inputs,thereby improving both richer feature encoding and enhanced model generalization.A feature filtering fusion module is developed to ensure the precise feature filtering and robust hybrid feature fusion.In response to the task-specific need for higher precision where false positives are very costly,traditional skip connections are replaced with the attention-guided feature reconstructing fusion module.Additionally,innovative data augmentation and balancing methods are proposed to counter data scarcity and distribution imbalance,further boosting the robustness and generalization of the model.With a comprehensive suite of evaluation metrics,extensive validations on four benchmark datasets(DRIVE,CHASEDB1,STARE,and HRF)and an SLO dataset(IOSTAR),demonstrate the proposed method’s superiority over both baseline and state-of-the-art models.Particularly the proposed method significantly outperforms the compared methods in cross-validation model generalization.
基金supported by the National Natural Science Foundation of China(Grant No.62001440).
文摘Quantum communication networks,such as quantum key distribution(QKD)networks,typically employ the measurement-resend mechanism between two users using quantum communication devices based on different quantum encoding types.To achieve direct communication between the devices with different quantum encoding types,in this paper,we propose encoding conversion schemes between the polarization bases(rectilinear,diagonal and circular bases)and the time-bin phase bases(two phase bases and time-bin basis)and design the quantum encoding converters.The theoretical analysis of the encoding conversion schemes is given in detail,and the basis correspondence of encoding conversion and the property of bit flip are revealed.The conversion relationship between polarization bases and time-bin phase bases can be easily selected by controlling a phase shifter.Since no optical switches are used in our scheme,the converter can be operated with high speed.The converters can also be modularized,which may be utilized to realize miniaturization in the future.
基金sponsored by the National Natural Science Foundation of China No.U24B201114,6247070859,62302114 and No.62172353Innovation Fund Program of the Engineering Research Center for Integration and Application of Digital Learning Technology of Ministry of Education No.1331007 and No.1311022Natural Science Foundation of Guangdong Province No.2024A1515010177.
文摘Blockchain,as a distributed ledger,inherently possesses tamper-resistant capabilities,creating a natural channel for covert communication.However,the immutable nature of data storage might introduce challenges to communication security.This study introduces a blockchain-based covert communication model utilizing dynamic Base-K encoding.The proposed encoding scheme utilizes the input address sequence to determine K to encode the secret message and determines the order of transactions based on K,thus ensuring effective concealment of the message.The dynamic encoding parameters enhance flexibility and address issues related to identical transaction amounts for the same secret message.Experimental results demonstrate that the proposed method maintains smooth communication and low susceptibility to tampering,achieving commendable concealment and embedding rates.
基金supported by the Fundamental Research Funds for the Central Universities(No.2024JBZX027)the National Natural Science Foundation of China(No.52375078).
文摘High-Speed Trains (HSTs) have emerged as a mainstream mode of transportation in China, owing to their exceptional safety and efficiency. Ensuring the reliable operation of HSTs is of paramount economic and societal importance. As critical rotating mechanical components of the transmission system, bearings make their fault diagnosis a topic of extensive attention. This paper provides a systematic review of image encoding-based bearing fault diagnosis methods tailored to the condition monitoring of HSTs. First, it categorizes the image encoding techniques applied in the field of bearing fault diagnosis. Then, a review of state-of-the-art studies has been presented, encompassing both monomodal image conversion and multimodal image fusion approaches. Finally, it highlights current challenges and proposes future research directions to advance intelligent fault diagnosis in HSTs, aiming to provide a valuable reference for researchers and engineers in the field of intelligent operation and maintenance.
基金financially supported by the Swedish Research Council(2022-04422_VR)。
文摘The Gaussian phase distribution approximation enables analysis of restricted diffusion encoded by general gradient waveforms but fails to account for the diffraction-like features that may occur for simple pore geometries.We investigate the range of validity of the approximation by random walk simulations of restricted diffusion in a cylinder using isotropic diffusion encoding sequences as well as conventional single gradient pulse pairs and oscillating gradient waveforms.The results show that clear deviations from the approximation may be observed at relative signal attenuations below 0.1 for onedimensional sequences with few oscillation periods.Increasing the encoding dimensionality and/or number of oscillations while extending the total duration of the waveform diminishes the non-Gaussian effects while preserving the low apparent diffusivities characteristic of restriction.
文摘Faulty-feeder detection in neutral point noneffectively grounded distribution networks consistently attracts research attention since it directly affects quality and safety of energy supply.Most modern research on faulty-feeder detection tends to apply more complex digital signal processing techniques and deeper neural networks in order to better extract and learn as many detailed characteristics as possible.However,these approaches may easily result in overfitting and high computational cost,which cannot meet requirements for detection accuracy and efficiency in practical applications.This paper proposes an innovative waveform encoding method and details a simple convolutional neural network(CNN)with one layer of convolution used for identification,which seeks to improve detection accuracy and efficiency simultaneously.First,sparse characteristics of waveforms are utilized to encode into compact vectors,and a waveform-vector matrix is generated.Second,to deduce waveform-vector matrix,a simple CNN with multi-scale filters and one layer of convolution is established.Finally,a methodology for faulty-feeder detection is proposed,and both detection accuracy and efficiency are considerably enhanced.Comparative studies have confirmed clear superiority of the developed method,which outperforms existing approaches in both detection accuracy and efficiency,thus highlighting its significant potential for application.
基金supported by the Collaborative Tackling Project of the Yangtze River Delta SciTech Innovation Community(Nos.2024CSJGG01503,2024CSJGG01500)Guangxi Key Research and Development Program(No.AB24010317)Jiangxi Provincial Key Laboratory of Electronic Data Control and Forensics(Jiangxi Police College)(No.2025JXJYKFJJ002).
文摘Multimodal sentiment analysis aims to understand emotions from text,speech,and video data.However,current methods often overlook the dominant role of text and suffer from feature loss during integration.Given the varying importance of each modality across different contexts,a central and pressing challenge in multimodal sentiment analysis lies in maximizing the use of rich intra-modal features while minimizing information loss during the fusion process.In response to these critical limitations,we propose a novel framework that integrates spatial position encoding and fusion embedding modules to address these issues.In our model,text is treated as the core modality,while speech and video features are selectively incorporated through a unique position-aware fusion process.The spatial position encoding strategy preserves the internal structural information of speech and visual modalities,enabling the model to capture localized intra-modal dependencies that are often overlooked.This design enhances the richness and discriminative power of the fused representation,enabling more accurate and context-aware sentiment prediction.Finally,we conduct comprehensive evaluations on two widely recognized standard datasets in the field—CMU-MOSI and CMU-MOSEI to validate the performance of the proposed model.The experimental results demonstrate that our model exhibits good performance and effectiveness for sentiment analysis tasks.
基金supported by the National Key Research and Development Project(Grant No.2023YFB4705300)the National Natural Science Foundation of China(NSFC)(Grant Nos.62205200 and 62375168)the Natural Science Foundation of Shanghai(Grant No.22ZR1431600)。
文摘Highly programmable shape morphing of 4D-printed micro/nanostructures is urgently desired for applications in robotics and intelligent systems.However,due to the lack of autonomous holistic strategies throughout the target shape input,optimal material distribution generation,and fabrication program output,4D nanoprinting that permits arbitrary shape morphing remains a challenging task for manual design.In this study,we report an autonomous inverse encoding strategy to decipher the genetic code for material property distributions that can guide the encoded modeling toward arbitrarily pre-programmed 4D shape morphing.By tuning the laser power of each voxel at the nanoscale,the genetic code can be spatially programmed and controllable shape morphing can be realized through the inverse encoding process.Using this strategy,the 4D-printed structures can be designed and accurately shift to the target morphing of arbitrarily hand-drawn lines under stimulation.Furthermore,as a proof-of-concept,a flexible fiber micromanipulator that can approach the target region through pre-programmed shape morphing is autonomously inversely encoded according to the localized spatial environment.This strategy may contribute to the modeling and arbitrary shape morphing of micro/nanostructures fabricated via 4D nanoprinting,leading to cutting-edge applications in microfluidics,micro-robotics,minimally invasive robotic surgery,and tissue engineering.
基金the National Natural Science Foundation of China(No.61861023)the Yunnan Fundamental Research Project(No.202301AT070452)。
文摘Sensitivity encoding(SENSE)is a parallel magnetic resonance imaging(MRI)reconstruction model by utilizing the sensitivity information of receiver coils to achieve image reconstruction.The existing SENSE-based reconstruction algorithms usually used nonadaptive sparsifying transforms,resulting in a limited reconstruction accuracy.Therefore,we proposed a new model for accurate parallel MRI reconstruction by combining the L0 norm regularization term based on the efficient sum of outer products dictionary learning(SOUPDIL)with the SENSE model,called SOUPDIL-SENSE.The SOUPDIL-SENSE model is mainly solved by utilizing the variable splitting and alternating direction method of multipliers techniques.The experimental results on four human datasets show that the proposed algorithm effectively promotes the image sparsity,eliminates the noise and artifacts of the reconstructed images,and improves the reconstruction accuracy.
