Sign language is a primary mode of communication for individuals with hearing impairments,conveying meaning through hand shapes and hand movements.Contrary to spoken or written languages,sign language relies on the re...Sign language is a primary mode of communication for individuals with hearing impairments,conveying meaning through hand shapes and hand movements.Contrary to spoken or written languages,sign language relies on the recognition and interpretation of hand gestures captured in video data.However,sign language datasets remain relatively limited compared to those of other languages,which hinders the training and performance of deep learning models.Additionally,the distinct word order of sign language,unlike that of spoken language,requires context-aware and natural sentence generation.To address these challenges,this study applies data augmentation techniques to build a Korean Sign Language dataset and train recognition models.Recognized words are then reconstructed into complete sentences.The sign recognition process uses OpenCV and MediaPipe to extract hand landmarks from sign language videos and analyzes hand position,orientation,and motion.The extracted features are converted into time-series data and fed into a Long Short-Term Memory(LSTM)model.The proposed recognition framework achieved an accuracy of up to 81.25%,while the sentence generation achieved an accuracy of up to 95%.The proposed approach is expected to be applicable not only to Korean Sign Language but also to other low-resource sign languages for recognition and translation tasks.展开更多
Deep neural networks have achieved excellent classification results on several computer vision benchmarks.This has led to the popularity of machine learning as a service,where trained algorithms are hosted on the clou...Deep neural networks have achieved excellent classification results on several computer vision benchmarks.This has led to the popularity of machine learning as a service,where trained algorithms are hosted on the cloud and inference can be obtained on real-world data.In most applications,it is important to compress the vision data due to the enormous bandwidth and memory requirements.Video codecs exploit spatial and temporal correlations to achieve high compression ratios,but they are computationally expensive.This work computes the motion fields between consecutive frames to facilitate the efficient classification of videos.However,contrary to the normal practice of reconstructing the full-resolution frames through motion compensation,this work proposes to infer the class label from the block-based computed motion fields directly.Motion fields are a richer and more complex representation of motion vectors,where each motion vector carries the magnitude and direction information.This approach has two advantages:the cost of motion compensation and video decoding is avoided,and the dimensions of the input signal are highly reduced.This results in a shallower network for classification.The neural network can be trained using motion vectors in two ways:complex representations and magnitude-direction pairs.The proposed work trains a convolutional neural network on the direction and magnitude tensors of the motion fields.Our experimental results show 20×faster convergence during training,reduced overfitting,and accelerated inference on a hand gesture recognition dataset compared to full-resolution and downsampled frames.We validate the proposed methodology on the HGds dataset,achieving a testing accuracy of 99.21%,on the HMDB51 dataset,achieving 82.54%accuracy,and on the UCF101 dataset,achieving 97.13%accuracy,outperforming state-of-the-art methods in computational efficiency.展开更多
This study focuses on the risks associated with the on-balance sheet recognition of data resources.At the legal level,disputes over ownership often arise due to unclear data property rights,while privacy protection,cy...This study focuses on the risks associated with the on-balance sheet recognition of data resources.At the legal level,disputes over ownership often arise due to unclear data property rights,while privacy protection,cybersecurity,and cross-border data flows create additional compliance challenges.In terms of recognition,the subjectivity of traditional valuation methods,the lack of active markets,and the rapid depreciation of data value caused by technological iteration hinder reliable measurement.With respect to disclosure,organizations face a dilemma between transparency and confidentiality.Collectively,these issues exacerbate audit risks.It is therefore imperative to establish an appropriate legal,accounting,and auditing framework to mitigate such risks and remove barriers to the proper recognition of data assets on balance sheets.展开更多
This research investigates the application of multisource data fusion using a Multi-Layer Perceptron (MLP) for Human Activity Recognition (HAR). The study integrates four distinct open-source datasets—WISDM, DaLiAc, ...This research investigates the application of multisource data fusion using a Multi-Layer Perceptron (MLP) for Human Activity Recognition (HAR). The study integrates four distinct open-source datasets—WISDM, DaLiAc, MotionSense, and PAMAP2—to develop a generalized MLP model for classifying six human activities. Performance analysis of the fused model for each dataset reveals accuracy rates of 95.83 for WISDM, 97 for DaLiAc, 94.65 for MotionSense, and 98.54 for PAMAP2. A comparative evaluation was conducted between the fused MLP model and the individual dataset models, with the latter tested on separate validation sets. The results indicate that the MLP model, trained on the fused dataset, exhibits superior performance relative to the models trained on individual datasets. This finding suggests that multisource data fusion significantly enhances the generalization and accuracy of HAR systems. The improved performance underscores the potential of integrating diverse data sources to create more robust and comprehensive models for activity recognition.展开更多
The consultation intention of emergency decision-makers in urban rail transit(URT)is input into the emergency knowledge base in the form of domain questions to obtain emergency decision support services.This approach ...The consultation intention of emergency decision-makers in urban rail transit(URT)is input into the emergency knowledge base in the form of domain questions to obtain emergency decision support services.This approach facilitates the rapid collection of complete knowledge and rules to form effective decisions.However,the current structured degree of the URT emergency knowledge base remains low,and the domain questions lack labeled datasets,resulting in a large deviation between the consultation outcomes and the intended objectives.To address this issue,this paper proposes a question intention recognition model for the URT emergency domain,leveraging knowledge graph(KG)and data enhancement technology.First,a structured storage of emergency cases and emergency plans is realized based on KG.Subsequently,a comprehensive question template is developed,and the labeled dataset of emergency domain questions in URT is generated through the KG.Lastly,data enhancement is applied by prompt learning and the NLP Chinese Data Augmentation(NLPCDA)tool,and the intention recognition model combining Generalized Auto-regression Pre-training for Language Understanding(XLNet)and Recurrent Convolutional Neural Network for Text Classification(TextRCNN)is constructed.Word embeddings are generated by XLNet,context information is further captured using Bidirectional Long Short-Term Memory Neural Network(BiLSTM),and salient features are extracted with Convolutional Neural Network(CNN).Experimental results demonstrate that the proposed model can enhance the clarity of classification and the identification of domain questions,thereby providing supportive knowledge for emergency decision-making in URT.展开更多
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
This study focuses on developing a deep learning model capable of recognizing vehicle brands and models,integrated with a law enforcement intelligence platform to overcome the limitations of existing license plate rec...This study focuses on developing a deep learning model capable of recognizing vehicle brands and models,integrated with a law enforcement intelligence platform to overcome the limitations of existing license plate recognition techniques—particularly in handling counterfeit,obscured,or absent plates.The research first entailed collecting,annotating,and classifying images of various vehiclemodels,leveraging image processing and feature extraction methodologies to train themodel on Microsoft Custom Vision.Experimental results indicate that,formost brands and models,the system achieves stable and relatively high performance in Precision,Recall,and Average Precision(AP).Furthermore,simulated tests involving illicit vehicles reveal that,even in cases of reassigned,concealed,or missing license plates,the model can rely on exterior body features to effectively identify vehicles,reducing dependence on plate-specific data.In practical law enforcement scenarios,these findings can accelerate investigations of stolen or forged plates and enhance overall accuracy.In conclusion,continued collection of vehicle images across broadermodel types,production years,and modification levels—along with refined annotation processes and parameter adjustment strategies—will further strengthen themethod’s applicability within law enforcement intelligence platforms,facilitating more precise and comprehensive vehicle recognition and control in real-world operations.展开更多
To support the process of grasping objects on a tabletop for the blind or robotic arm,it is necessary to address fundamental computer vision tasks,such as detecting,recognizing,and locating objects in space,and determ...To support the process of grasping objects on a tabletop for the blind or robotic arm,it is necessary to address fundamental computer vision tasks,such as detecting,recognizing,and locating objects in space,and determining the position of the grasping information.These results can then be used to guide the visually impaired or to execute grasping tasks with a robotic arm.In this paper,we collected,annotated,and published the benchmark TQUGraspingObject dataset for testing,validation,and evaluation of deep learning(DL)models for detecting,recognizing,and localizing grasping objects in 2D and 3D space,especially 3D point cloud data.Our dataset is collected in a shared room,with common everyday objects placed on the tabletop in jumbled positions by Intel RealSense D435(IR-D435).This dataset includes more than 63k RGB-D pairs and related data such as normalized 3D object point cloud,3D object point cloud segmented,coordinate system normalizationmatrix,3D object point cloud normalized,and hand pose for grasping each object.At the same time,we also conducted experiments on fourDL networks with the best performance:SSD-MobileNetV3,ResNet50-Transformer,ResNet101-Transformer,and YOLOv12.The results present that YOLOv12 has the most suitable results in detecting and recognizing objects in images.All data,annotations,toolkit,source code,point cloud data,and results are publicly available on our project website:https://github.com/HuaTThanhIT2327Tqu/datasetv2.展开更多
In order to effectively conduct emotion recognition from spontaneous, non-prototypical and unsegmented speech so as to create a more natural human-machine interaction; a novel speech emotion recognition algorithm base...In order to effectively conduct emotion recognition from spontaneous, non-prototypical and unsegmented speech so as to create a more natural human-machine interaction; a novel speech emotion recognition algorithm based on the combination of the emotional data field （EDF） and the ant colony search （ACS） strategy, called the EDF-ACS algorithm, is proposed. More specifically, the inter- relationship among the turn-based acoustic feature vectors of different labels are established by using the potential function in the EDF. To perform the spontaneous speech emotion recognition, the artificial colony is used to mimic the turn- based acoustic feature vectors. Then, the canonical ACS strategy is used to investigate the movement direction of each artificial ant in the EDF, which is regarded as the emotional label of the corresponding turn-based acoustic feature vector. The proposed EDF-ACS algorithm is evaluated on the continueous audio）＇visual emotion challenge （AVEC） 2012 dataset, which contains the spontaneous, non-prototypical and unsegmented speech emotion data. The experimental results show that the proposed EDF-ACS algorithm outperforms the existing state-of-the-art algorithm in turn-based speech emotion recognition.展开更多
Edge detection and enhancement techniques are commonly used in recognizing the edge of geologic bodies using potential field data. We present a new edge recognition technology based on the normalized vertical derivati...Edge detection and enhancement techniques are commonly used in recognizing the edge of geologic bodies using potential field data. We present a new edge recognition technology based on the normalized vertical derivative of the total horizontal derivative which has the functions of both edge detection and enhancement techniques. First, we calculate the total horizontal derivative （THDR） of the potential-field data and then compute the n-order vertical derivative （VDRn） of the THDR. For the n-order vertical derivative, the peak value of total horizontal derivative （PTHDR） is obtained using a threshold value greater than 0. This PTHDR can be used for edge detection. Second, the PTHDR value is divided by the total horizontal derivative and normalized by the maximum value. Finally, we used different kinds of numerical models to verify the effectiveness and reliability of the new edge recognition technology.展开更多
The coal-rock interface recognition method based on multi-sensor data fusiontechnique is put forward because of the localization of single type sensor recognition method. Themeasuring theory based on multi-sensor data...The coal-rock interface recognition method based on multi-sensor data fusiontechnique is put forward because of the localization of single type sensor recognition method. Themeasuring theory based on multi-sensor data fusion technique is analyzed, and hereby the testplatform of recognition system is manufactured. The advantage of data fusion with the fuzzy neuralnetwork (FNN) technique has been probed. The two-level FNN is constructed and data fusion is carriedout. The experiments show that in various conditions the method can always acquire a much higherrecognition rate than normal ones.展开更多
Communication behavior recognition is an issue with increasingly importance in the antiterrorism and national defense area.However,the sensing data obtained in actual environment is often not sufficient to accurately ...Communication behavior recognition is an issue with increasingly importance in the antiterrorism and national defense area.However,the sensing data obtained in actual environment is often not sufficient to accurately analyze the communication behavior.Traditional means can hardly utilize the scarce and crude spectrum sensing data captured in a real scene.Thus,communication behavior recognition using raw sensing data under smallsample condition has become a new challenge.In this paper,a data enhanced communication behavior recognition(DECBR)scheme is proposed to meet this challenge.Firstly,a preprocessing method is designed to make the raw spectrum data suitable for the proposed scheme.Then,an adaptive convolutional neural network structure is exploited to carry out communication behavior recognition.Moreover,DCGAN is applied to support data enhancement,which realize communication behavior recognition under small-sample condition.Finally,the scheme is verified by experiments under different data size.The results show that the DECBR scheme can greatly improve the accuracy and efficiency of behavior recognition under smallsample condition.展开更多
In computational physics proton transfer phenomena could be viewed as pattern classification problems based on a set of input features allowing classification of the proton motion into two categories： transfer ＇occu...In computational physics proton transfer phenomena could be viewed as pattern classification problems based on a set of input features allowing classification of the proton motion into two categories： transfer ＇occurred＇ and transfer ＇not occurred＇. The goal of this paper is to evaluate the use of artificial neural networks in the classification of proton transfer events, based on the feed-forward back propagation neural network, used as a classifier to distinguish between the two transfer cases. In this paper, we use a new developed data mining and pattern recognition tool for automating, controlling, and drawing charts of the output data of an Empirical Valence Bond existing code. The study analyzes the need for pattern recognition in aqueous proton transfer processes and how the learning approach in error back propagation （multilayer perceptron algorithms） could be satisfactorily employed in the present case. We present a tool for pattern recognition and validate the code including a real physical case study. The results of applying the artificial neural networks methodology to crowd patterns based upon selected physical properties （e.g., temperature, density） show the abilities of the network to learn proton transfer patterns corresponding to properties of the aqueous environments, which is in turn proved to be fully compatible with previous proton transfer studies.展开更多
The ability of accurate and scalable mobile device recognition is critically important for mobile network operators and ISPs to understand their customers' behaviours and enhance their user experience.In this pape...The ability of accurate and scalable mobile device recognition is critically important for mobile network operators and ISPs to understand their customers' behaviours and enhance their user experience.In this paper,we propose a novel method for mobile device model recognition by using statistical information derived from large amounts of mobile network traffic data.Specifically,we create a Jaccardbased coefficient measure method to identify a proper keyword representing each mobile device model from massive unstructured textual HTTP access logs.To handle the large amount of traffic data generated from large mobile networks,this method is designed as a set of parallel algorithms,and is implemented through the MapReduce framework which is a distributed parallel programming model with proven low-cost and high-efficiency features.Evaluations using real data sets show that our method can accurately recognise mobile client models while meeting the scalability and producer-independency requirements of large mobile network operators.Results show that a 91.5% accuracy rate is achieved for recognising mobile client models from 2 billion records,which is dramatically higher than existing solutions.展开更多
In recent years,deep learning-based signal recognition technology has gained attention and emerged as an important approach for safeguarding the electromagnetic environment.However,training deep learning-based classif...In recent years,deep learning-based signal recognition technology has gained attention and emerged as an important approach for safeguarding the electromagnetic environment.However,training deep learning-based classifiers on large signal datasets with redundant samples requires significant memory and high costs.This paper proposes a support databased core-set selection method(SD)for signal recognition,aiming to screen a representative subset that approximates the large signal dataset.Specifically,this subset can be identified by employing the labeled information during the early stages of model training,as some training samples are labeled as supporting data frequently.This support data is crucial for model training and can be found using a border sample selector.Simulation results demonstrate that the SD method minimizes the impact on model recognition performance while reducing the dataset size,and outperforms five other state-of-the-art core-set selection methods when the fraction of training sample kept is less than or equal to 0.3 on the RML2016.04C dataset or 0.5 on the RML22 dataset.The SD method is particularly helpful for signal recognition tasks with limited memory and computing resources.展开更多
Automatic modulation recognition(AMR)of radiation source signals is a research focus in the field of cognitive radio.However,the AMR of radiation source signals at low SNRs still faces a great challenge.Therefore,the ...Automatic modulation recognition(AMR)of radiation source signals is a research focus in the field of cognitive radio.However,the AMR of radiation source signals at low SNRs still faces a great challenge.Therefore,the AMR method of radiation source signals based on two-dimensional data matrix and improved residual neural network is proposed in this paper.First,the time series of the radiation source signals are reconstructed into two-dimensional data matrix,which greatly simplifies the signal preprocessing process.Second,the depthwise convolution and large-size convolutional kernels based residual neural network(DLRNet)is proposed to improve the feature extraction capability of the AMR model.Finally,the model performs feature extraction and classification on the two-dimensional data matrix to obtain the recognition vector that represents the signal modulation type.Theoretical analysis and simulation results show that the AMR method based on two-dimensional data matrix and improved residual network can significantly improve the accuracy of the AMR method.The recognition accuracy of the proposed method maintains a high level greater than 90% even at -14 dB SNR.展开更多
A novel outlier recognition method in surveying data is presented based on Shannon information entropy. The probability distribution of surveying data does not need to be known or hypothesized in this method, and it i...A novel outlier recognition method in surveying data is presented based on Shannon information entropy. The probability distribution of surveying data does not need to be known or hypothesized in this method, and it is not only accurate but also convenient to calculate in this method compared with statistical recognition method.展开更多
With the rapid development of information technology,the electronifi-cation of medical records has gradually become a trend.In China,the population base is huge and the supporting medical institutions are numerous,so ...With the rapid development of information technology,the electronifi-cation of medical records has gradually become a trend.In China,the population base is huge and the supporting medical institutions are numerous,so this reality drives the conversion of paper medical records to electronic medical records.Electronic medical records are the basis for establishing a smart hospital and an important guarantee for achieving medical intelligence,and the massive amount of electronic medical record data is also an important data set for conducting research in the medical field.However,electronic medical records contain a large amount of private patient information,which must be desensitized before they are used as open resources.Therefore,to solve the above problems,data masking for Chinese electronic medical records with named entity recognition is proposed in this paper.Firstly,the text is vectorized to satisfy the required format of the model input.Secondly,since the input sentences may have a long or short length and the relationship between sentences in context is not negligible.To this end,a neural network model for named entity recognition based on bidirectional long short-term memory(BiLSTM)with conditional random fields(CRF)is constructed.Finally,the data masking operation is performed based on the named entity recog-nition results,mainly using regular expression filtering encryption and principal component analysis(PCA)word vector compression and replacement.In addi-tion,comparison experiments with the hidden markov model(HMM)model,LSTM-CRF model,and BiLSTM model are conducted in this paper.The experi-mental results show that the method used in this paper achieves 92.72%Accuracy,92.30%Recall,and 92.51%F1_score,which has higher accuracy compared with other models.展开更多
Data augmentation methods are often used to address data scarcity in natural language processing(NLP).However,token-label misalignment,which refers to situations where tokens are matched with incorrect entity labels i...Data augmentation methods are often used to address data scarcity in natural language processing(NLP).However,token-label misalignment,which refers to situations where tokens are matched with incorrect entity labels in the augmented sentences,hinders the data augmentation methods from achieving high scores in token-level tasks like named entity recognition(NER).In this paper,we propose embedded prompt tuning(EPT)as a novel data augmentation approach to low-resource NER.To address the problem of token-label misalignment,we implicitly embed NER labels as prompt into the hidden layer of pre-trained language model,and therefore entity tokens masked can be predicted by the finetuned EPT.Hence,EPT can generate high-quality and high-diverse data with various entities,which improves performance of NER.As datasets of cross-domain NER are available,we also explore NER domain adaption with EPT.The experimental results show that EPT achieves substantial improvement over the baseline methods on low-resource NER tasks.展开更多
基金supported by the Institute of Information&Communications Technoljogy Planning&Evaluation(IITP)-Innovative Human Resource Development for Local Intellectualization Program grant funded by the Korea government(MSIT)(IITP-2026-RS-2022-00156334,50%)the National Research Foundation of Korea(NRF)grant funded by the Korea government(MSIT)(No.2021R1C1C2011105,50%).
文摘Sign language is a primary mode of communication for individuals with hearing impairments,conveying meaning through hand shapes and hand movements.Contrary to spoken or written languages,sign language relies on the recognition and interpretation of hand gestures captured in video data.However,sign language datasets remain relatively limited compared to those of other languages,which hinders the training and performance of deep learning models.Additionally,the distinct word order of sign language,unlike that of spoken language,requires context-aware and natural sentence generation.To address these challenges,this study applies data augmentation techniques to build a Korean Sign Language dataset and train recognition models.Recognized words are then reconstructed into complete sentences.The sign recognition process uses OpenCV and MediaPipe to extract hand landmarks from sign language videos and analyzes hand position,orientation,and motion.The extracted features are converted into time-series data and fed into a Long Short-Term Memory(LSTM)model.The proposed recognition framework achieved an accuracy of up to 81.25%,while the sentence generation achieved an accuracy of up to 95%.The proposed approach is expected to be applicable not only to Korean Sign Language but also to other low-resource sign languages for recognition and translation tasks.
基金Supported by Princess Nourah bint Abdulrahman University Researchers Supporting Project number(PNURSP2025R896).
文摘Deep neural networks have achieved excellent classification results on several computer vision benchmarks.This has led to the popularity of machine learning as a service,where trained algorithms are hosted on the cloud and inference can be obtained on real-world data.In most applications,it is important to compress the vision data due to the enormous bandwidth and memory requirements.Video codecs exploit spatial and temporal correlations to achieve high compression ratios,but they are computationally expensive.This work computes the motion fields between consecutive frames to facilitate the efficient classification of videos.However,contrary to the normal practice of reconstructing the full-resolution frames through motion compensation,this work proposes to infer the class label from the block-based computed motion fields directly.Motion fields are a richer and more complex representation of motion vectors,where each motion vector carries the magnitude and direction information.This approach has two advantages:the cost of motion compensation and video decoding is avoided,and the dimensions of the input signal are highly reduced.This results in a shallower network for classification.The neural network can be trained using motion vectors in two ways:complex representations and magnitude-direction pairs.The proposed work trains a convolutional neural network on the direction and magnitude tensors of the motion fields.Our experimental results show 20×faster convergence during training,reduced overfitting,and accelerated inference on a hand gesture recognition dataset compared to full-resolution and downsampled frames.We validate the proposed methodology on the HGds dataset,achieving a testing accuracy of 99.21%,on the HMDB51 dataset,achieving 82.54%accuracy,and on the UCF101 dataset,achieving 97.13%accuracy,outperforming state-of-the-art methods in computational efficiency.
文摘This study focuses on the risks associated with the on-balance sheet recognition of data resources.At the legal level,disputes over ownership often arise due to unclear data property rights,while privacy protection,cybersecurity,and cross-border data flows create additional compliance challenges.In terms of recognition,the subjectivity of traditional valuation methods,the lack of active markets,and the rapid depreciation of data value caused by technological iteration hinder reliable measurement.With respect to disclosure,organizations face a dilemma between transparency and confidentiality.Collectively,these issues exacerbate audit risks.It is therefore imperative to establish an appropriate legal,accounting,and auditing framework to mitigate such risks and remove barriers to the proper recognition of data assets on balance sheets.
基金supported by the Royal Golden Jubilee(RGJ)Ph.D.Programme(Grant No.PHD/0079/2561)through the National Research Council of Thailand(NRCT)and Thailand Research Fund(TRF).
文摘This research investigates the application of multisource data fusion using a Multi-Layer Perceptron (MLP) for Human Activity Recognition (HAR). The study integrates four distinct open-source datasets—WISDM, DaLiAc, MotionSense, and PAMAP2—to develop a generalized MLP model for classifying six human activities. Performance analysis of the fused model for each dataset reveals accuracy rates of 95.83 for WISDM, 97 for DaLiAc, 94.65 for MotionSense, and 98.54 for PAMAP2. A comparative evaluation was conducted between the fused MLP model and the individual dataset models, with the latter tested on separate validation sets. The results indicate that the MLP model, trained on the fused dataset, exhibits superior performance relative to the models trained on individual datasets. This finding suggests that multisource data fusion significantly enhances the generalization and accuracy of HAR systems. The improved performance underscores the potential of integrating diverse data sources to create more robust and comprehensive models for activity recognition.
基金supported in part by the National Natural Science Foundation of China.The funding numbers 62433005,62272036,62132003,and 62173167.
文摘The consultation intention of emergency decision-makers in urban rail transit(URT)is input into the emergency knowledge base in the form of domain questions to obtain emergency decision support services.This approach facilitates the rapid collection of complete knowledge and rules to form effective decisions.However,the current structured degree of the URT emergency knowledge base remains low,and the domain questions lack labeled datasets,resulting in a large deviation between the consultation outcomes and the intended objectives.To address this issue,this paper proposes a question intention recognition model for the URT emergency domain,leveraging knowledge graph(KG)and data enhancement technology.First,a structured storage of emergency cases and emergency plans is realized based on KG.Subsequently,a comprehensive question template is developed,and the labeled dataset of emergency domain questions in URT is generated through the KG.Lastly,data enhancement is applied by prompt learning and the NLP Chinese Data Augmentation(NLPCDA)tool,and the intention recognition model combining Generalized Auto-regression Pre-training for Language Understanding(XLNet)and Recurrent Convolutional Neural Network for Text Classification(TextRCNN)is constructed.Word embeddings are generated by XLNet,context information is further captured using Bidirectional Long Short-Term Memory Neural Network(BiLSTM),and salient features are extracted with Convolutional Neural Network(CNN).Experimental results demonstrate that the proposed model can enhance the clarity of classification and the identification of domain questions,thereby providing supportive knowledge for emergency decision-making in URT.
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the National Science and Technology Council,Taiwan,for financially supporting this research(grant No.NSTC 114-2221-E-018-003)the Ministry of Education’s Teaching Practice Research Program,Taiwan(PSK1142780).
文摘This study focuses on developing a deep learning model capable of recognizing vehicle brands and models,integrated with a law enforcement intelligence platform to overcome the limitations of existing license plate recognition techniques—particularly in handling counterfeit,obscured,or absent plates.The research first entailed collecting,annotating,and classifying images of various vehiclemodels,leveraging image processing and feature extraction methodologies to train themodel on Microsoft Custom Vision.Experimental results indicate that,formost brands and models,the system achieves stable and relatively high performance in Precision,Recall,and Average Precision(AP).Furthermore,simulated tests involving illicit vehicles reveal that,even in cases of reassigned,concealed,or missing license plates,the model can rely on exterior body features to effectively identify vehicles,reducing dependence on plate-specific data.In practical law enforcement scenarios,these findings can accelerate investigations of stolen or forged plates and enhance overall accuracy.In conclusion,continued collection of vehicle images across broadermodel types,production years,and modification levels—along with refined annotation processes and parameter adjustment strategies—will further strengthen themethod’s applicability within law enforcement intelligence platforms,facilitating more precise and comprehensive vehicle recognition and control in real-world operations.
文摘To support the process of grasping objects on a tabletop for the blind or robotic arm,it is necessary to address fundamental computer vision tasks,such as detecting,recognizing,and locating objects in space,and determining the position of the grasping information.These results can then be used to guide the visually impaired or to execute grasping tasks with a robotic arm.In this paper,we collected,annotated,and published the benchmark TQUGraspingObject dataset for testing,validation,and evaluation of deep learning(DL)models for detecting,recognizing,and localizing grasping objects in 2D and 3D space,especially 3D point cloud data.Our dataset is collected in a shared room,with common everyday objects placed on the tabletop in jumbled positions by Intel RealSense D435(IR-D435).This dataset includes more than 63k RGB-D pairs and related data such as normalized 3D object point cloud,3D object point cloud segmented,coordinate system normalizationmatrix,3D object point cloud normalized,and hand pose for grasping each object.At the same time,we also conducted experiments on fourDL networks with the best performance:SSD-MobileNetV3,ResNet50-Transformer,ResNet101-Transformer,and YOLOv12.The results present that YOLOv12 has the most suitable results in detecting and recognizing objects in images.All data,annotations,toolkit,source code,point cloud data,and results are publicly available on our project website:https://github.com/HuaTThanhIT2327Tqu/datasetv2.
基金The National Natural Science Foundation of China(No.61231002,61273266,61571106)the Foundation of the Department of Science and Technology of Guizhou Province(No.[2015]7637)
文摘In order to effectively conduct emotion recognition from spontaneous, non-prototypical and unsegmented speech so as to create a more natural human-machine interaction; a novel speech emotion recognition algorithm based on the combination of the emotional data field （EDF） and the ant colony search （ACS） strategy, called the EDF-ACS algorithm, is proposed. More specifically, the inter- relationship among the turn-based acoustic feature vectors of different labels are established by using the potential function in the EDF. To perform the spontaneous speech emotion recognition, the artificial colony is used to mimic the turn- based acoustic feature vectors. Then, the canonical ACS strategy is used to investigate the movement direction of each artificial ant in the EDF, which is regarded as the emotional label of the corresponding turn-based acoustic feature vector. The proposed EDF-ACS algorithm is evaluated on the continueous audio）＇visual emotion challenge （AVEC） 2012 dataset, which contains the spontaneous, non-prototypical and unsegmented speech emotion data. The experimental results show that the proposed EDF-ACS algorithm outperforms the existing state-of-the-art algorithm in turn-based speech emotion recognition.
基金supported by the National Science and Technology Major Projects (2008ZX05025)the Project of National Oil and Gas Resources Strategic Constituency Survey and Evaluation of the Ministry of Land and Resources,China (XQ-2007-05)
文摘Edge detection and enhancement techniques are commonly used in recognizing the edge of geologic bodies using potential field data. We present a new edge recognition technology based on the normalized vertical derivative of the total horizontal derivative which has the functions of both edge detection and enhancement techniques. First, we calculate the total horizontal derivative （THDR） of the potential-field data and then compute the n-order vertical derivative （VDRn） of the THDR. For the n-order vertical derivative, the peak value of total horizontal derivative （PTHDR） is obtained using a threshold value greater than 0. This PTHDR can be used for edge detection. Second, the PTHDR value is divided by the total horizontal derivative and normalized by the maximum value. Finally, we used different kinds of numerical models to verify the effectiveness and reliability of the new edge recognition technology.
基金This project is supported by Provincial Youth Science Foundation of Shanxi China (No.20011020)National Natural Science Foundation of China (No.59975064).
文摘The coal-rock interface recognition method based on multi-sensor data fusiontechnique is put forward because of the localization of single type sensor recognition method. Themeasuring theory based on multi-sensor data fusion technique is analyzed, and hereby the testplatform of recognition system is manufactured. The advantage of data fusion with the fuzzy neuralnetwork (FNN) technique has been probed. The two-level FNN is constructed and data fusion is carriedout. The experiments show that in various conditions the method can always acquire a much higherrecognition rate than normal ones.
基金supported by the National Natural Science Foundation of China(No.61971439 and No.61702543)the Natural Science Foundation of the Jiangsu Province of China(No.BK20191329)+1 种基金the China Postdoctoral Science Foundation Project(No.2019T120987)the Startup Foundation for Introducing Talent of NUIST(No.2020r100).
文摘Communication behavior recognition is an issue with increasingly importance in the antiterrorism and national defense area.However,the sensing data obtained in actual environment is often not sufficient to accurately analyze the communication behavior.Traditional means can hardly utilize the scarce and crude spectrum sensing data captured in a real scene.Thus,communication behavior recognition using raw sensing data under smallsample condition has become a new challenge.In this paper,a data enhanced communication behavior recognition(DECBR)scheme is proposed to meet this challenge.Firstly,a preprocessing method is designed to make the raw spectrum data suitable for the proposed scheme.Then,an adaptive convolutional neural network structure is exploited to carry out communication behavior recognition.Moreover,DCGAN is applied to support data enhancement,which realize communication behavior recognition under small-sample condition.Finally,the scheme is verified by experiments under different data size.The results show that the DECBR scheme can greatly improve the accuracy and efficiency of behavior recognition under smallsample condition.
基金Dr. Steve Jones, Scientific Advisor of the Canon Foundation for Scientific Research (7200 The Quorum, Oxford Business Park, Oxford OX4 2JZ, England). Canon Foundation for Scientific Research funded the UPC 2013 tuition fees of the corresponding author during her writing this article
文摘In computational physics proton transfer phenomena could be viewed as pattern classification problems based on a set of input features allowing classification of the proton motion into two categories： transfer ＇occurred＇ and transfer ＇not occurred＇. The goal of this paper is to evaluate the use of artificial neural networks in the classification of proton transfer events, based on the feed-forward back propagation neural network, used as a classifier to distinguish between the two transfer cases. In this paper, we use a new developed data mining and pattern recognition tool for automating, controlling, and drawing charts of the output data of an Empirical Valence Bond existing code. The study analyzes the need for pattern recognition in aqueous proton transfer processes and how the learning approach in error back propagation （multilayer perceptron algorithms） could be satisfactorily employed in the present case. We present a tool for pattern recognition and validate the code including a real physical case study. The results of applying the artificial neural networks methodology to crowd patterns based upon selected physical properties （e.g., temperature, density） show the abilities of the network to learn proton transfer patterns corresponding to properties of the aqueous environments, which is in turn proved to be fully compatible with previous proton transfer studies.
基金supported in part by the National Natural Science Foundation of China under Grant No.61072061the National Science and Technology Major Projects under Grant No.2012ZX03002008the Fundamental Research Funds for the Central Universities under Grant No.2012RC0121
文摘The ability of accurate and scalable mobile device recognition is critically important for mobile network operators and ISPs to understand their customers' behaviours and enhance their user experience.In this paper,we propose a novel method for mobile device model recognition by using statistical information derived from large amounts of mobile network traffic data.Specifically,we create a Jaccardbased coefficient measure method to identify a proper keyword representing each mobile device model from massive unstructured textual HTTP access logs.To handle the large amount of traffic data generated from large mobile networks,this method is designed as a set of parallel algorithms,and is implemented through the MapReduce framework which is a distributed parallel programming model with proven low-cost and high-efficiency features.Evaluations using real data sets show that our method can accurately recognise mobile client models while meeting the scalability and producer-independency requirements of large mobile network operators.Results show that a 91.5% accuracy rate is achieved for recognising mobile client models from 2 billion records,which is dramatically higher than existing solutions.
基金supported by National Natural Science Foundation of China(62371098)Natural Science Foundation of Sichuan Province(2023NSFSC1422)+1 种基金National Key Research and Development Program of China(2021YFB2900404)Central Universities of South west Minzu University(ZYN2022032).
文摘In recent years,deep learning-based signal recognition technology has gained attention and emerged as an important approach for safeguarding the electromagnetic environment.However,training deep learning-based classifiers on large signal datasets with redundant samples requires significant memory and high costs.This paper proposes a support databased core-set selection method(SD)for signal recognition,aiming to screen a representative subset that approximates the large signal dataset.Specifically,this subset can be identified by employing the labeled information during the early stages of model training,as some training samples are labeled as supporting data frequently.This support data is crucial for model training and can be found using a border sample selector.Simulation results demonstrate that the SD method minimizes the impact on model recognition performance while reducing the dataset size,and outperforms five other state-of-the-art core-set selection methods when the fraction of training sample kept is less than or equal to 0.3 on the RML2016.04C dataset or 0.5 on the RML22 dataset.The SD method is particularly helpful for signal recognition tasks with limited memory and computing resources.
基金National Natural Science Foundation of China under Grant No.61973037China Postdoctoral Science Foundation under Grant No.2022M720419。
文摘Automatic modulation recognition(AMR)of radiation source signals is a research focus in the field of cognitive radio.However,the AMR of radiation source signals at low SNRs still faces a great challenge.Therefore,the AMR method of radiation source signals based on two-dimensional data matrix and improved residual neural network is proposed in this paper.First,the time series of the radiation source signals are reconstructed into two-dimensional data matrix,which greatly simplifies the signal preprocessing process.Second,the depthwise convolution and large-size convolutional kernels based residual neural network(DLRNet)is proposed to improve the feature extraction capability of the AMR model.Finally,the model performs feature extraction and classification on the two-dimensional data matrix to obtain the recognition vector that represents the signal modulation type.Theoretical analysis and simulation results show that the AMR method based on two-dimensional data matrix and improved residual network can significantly improve the accuracy of the AMR method.The recognition accuracy of the proposed method maintains a high level greater than 90% even at -14 dB SNR.
文摘A novel outlier recognition method in surveying data is presented based on Shannon information entropy. The probability distribution of surveying data does not need to be known or hypothesized in this method, and it is not only accurate but also convenient to calculate in this method compared with statistical recognition method.
基金This research was supported by the National Natural Science Foundation of China under Grant(No.42050102)the Postgraduate Education Reform Project of Jiangsu Province under Grant(No.SJCX22_0343)Also,this research was supported by Dou Wanchun Expert Workstation of Yunnan Province(No.202205AF150013).
文摘With the rapid development of information technology,the electronifi-cation of medical records has gradually become a trend.In China,the population base is huge and the supporting medical institutions are numerous,so this reality drives the conversion of paper medical records to electronic medical records.Electronic medical records are the basis for establishing a smart hospital and an important guarantee for achieving medical intelligence,and the massive amount of electronic medical record data is also an important data set for conducting research in the medical field.However,electronic medical records contain a large amount of private patient information,which must be desensitized before they are used as open resources.Therefore,to solve the above problems,data masking for Chinese electronic medical records with named entity recognition is proposed in this paper.Firstly,the text is vectorized to satisfy the required format of the model input.Secondly,since the input sentences may have a long or short length and the relationship between sentences in context is not negligible.To this end,a neural network model for named entity recognition based on bidirectional long short-term memory(BiLSTM)with conditional random fields(CRF)is constructed.Finally,the data masking operation is performed based on the named entity recog-nition results,mainly using regular expression filtering encryption and principal component analysis(PCA)word vector compression and replacement.In addi-tion,comparison experiments with the hidden markov model(HMM)model,LSTM-CRF model,and BiLSTM model are conducted in this paper.The experi-mental results show that the method used in this paper achieves 92.72%Accuracy,92.30%Recall,and 92.51%F1_score,which has higher accuracy compared with other models.
文摘Data augmentation methods are often used to address data scarcity in natural language processing(NLP).However,token-label misalignment,which refers to situations where tokens are matched with incorrect entity labels in the augmented sentences,hinders the data augmentation methods from achieving high scores in token-level tasks like named entity recognition(NER).In this paper,we propose embedded prompt tuning(EPT)as a novel data augmentation approach to low-resource NER.To address the problem of token-label misalignment,we implicitly embed NER labels as prompt into the hidden layer of pre-trained language model,and therefore entity tokens masked can be predicted by the finetuned EPT.Hence,EPT can generate high-quality and high-diverse data with various entities,which improves performance of NER.As datasets of cross-domain NER are available,we also explore NER domain adaption with EPT.The experimental results show that EPT achieves substantial improvement over the baseline methods on low-resource NER tasks.
