Purpose:With more and more digital collections of various information resources becoming available,also increasing is the challenge of assigning subject index terms and classes from quality knowledge organization syst...Purpose:With more and more digital collections of various information resources becoming available,also increasing is the challenge of assigning subject index terms and classes from quality knowledge organization systems.While the ultimate purpose is to understand the value of automatically produced Dewey Decimal Classification(DDC)classes for Swedish digital collections,the paper aims to evaluate the performance of six machine learning algorithms as well as a string-matching algorithm based on characteristics of DDC.Design/methodology/approach:State-of-the-art machine learning algorithms require at least 1,000 training examples per class.The complete data set at the time of research involved 143,838 records which had to be reduced to top three hierarchical levels of DDC in order to provide sufficient training data(totaling 802 classes in the training and testing sample,out of 14,413 classes at all levels).Findings:Evaluation shows that Support Vector Machine with linear kernel outperforms other machine learning algorithms as well as the string-matching algorithm on average;the string-matching algorithm outperforms machine learning for specific classes when characteristics of DDC are most suitable for the task.Word embeddings combined with different types of neural networks(simple linear network,standard neural network,1 D convolutional neural network,and recurrent neural network)produced worse results than Support Vector Machine,but reach close results,with the benefit of a smaller representation size.Impact of features in machine learning shows that using keywords or combining titles and keywords gives better results than using only titles as input.Stemming only marginally improves the results.Removed stop-words reduced accuracy in most cases,while removing less frequent words increased it marginally.The greatest impact is produced by the number of training examples:81.90%accuracy on the training set is achieved when at least 1,000 records per class are available in the training set,and 66.13%when too few records(often less than A Comparison of Approaches100 per class)on which to train are available—and these hold only for top 3 hierarchical levels(803 instead of 14,413 classes).Research limitations:Having to reduce the number of hierarchical levels to top three levels of DDC because of the lack of training data for all classes,skews the results so that they work in experimental conditions but barely for end users in operational retrieval systems.Practical implications:In conclusion,for operative information retrieval systems applying purely automatic DDC does not work,either using machine learning(because of the lack of training data for the large number of DDC classes)or using string-matching algorithm(because DDC characteristics perform well for automatic classification only in a small number of classes).Over time,more training examples may become available,and DDC may be enriched with synonyms in order to enhance accuracy of automatic classification which may also benefit information retrieval performance based on DDC.In order for quality information services to reach the objective of highest possible precision and recall,automatic classification should never be implemented on its own;instead,machine-aided indexing that combines the efficiency of automatic suggestions with quality of human decisions at the final stage should be the way for the future.Originality/value:The study explored machine learning on a large classification system of over 14,000 classes which is used in operational information retrieval systems.Due to lack of sufficient training data across the entire set of classes,an approach complementing machine learning,that of string matching,was applied.This combination should be explored further since it provides the potential for real-life applications with large target classification systems.展开更多
To promote reliable and secure communications in the cognitive radio network,the automatic modulation classification algorithms have been mainly proposed to estimate a single modulation.In this paper,we address the cl...To promote reliable and secure communications in the cognitive radio network,the automatic modulation classification algorithms have been mainly proposed to estimate a single modulation.In this paper,we address the classification of superimposed modulations dedicated to 5G multipleinput multiple-output(MIMO)two-way cognitive relay network in realistic channels modeled with Nakagami-m distribution.Our purpose consists of classifying pairs of users modulations from superimposed signals.To achieve this goal,we apply the higher-order statistics in conjunction with the Multi-BoostAB classifier.We use several efficiency metrics including the true positive(TP)rate,false positive(FP)rate,precision,recall,F-Measure and receiver operating characteristic(ROC)area in order to evaluate the performance of the proposed algorithm in terms of correct superimposed modulations classification.Computer simulations prove that our proposal allows obtaining a good probability of classification for ten superimposed modulations at a low signal-to-noise ratio,including the worst case(i.e.,m=0.5),where the fading distribution follows a one-sided Gaussian distribution.We also carry out a comparative study between our proposal usingMultiBoostAB classifier with the decision tree(J48)classifier.Simulation results show that the performance of MultiBoostAB on the superimposed modulations classifications outperforms the one of J48 classifier.In addition,we study the impact of the symbols number,path loss exponent and relay position on the performance of the proposed automatic classification superimposed modulations in terms of probability of correct classification.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
Due to the limited computational capability and the diversity of the Internet of Things devices working in different environment,we consider fewshot learning-based automatic modulation classification(AMC)to improve it...Due to the limited computational capability and the diversity of the Internet of Things devices working in different environment,we consider fewshot learning-based automatic modulation classification(AMC)to improve its reliability.A data enhancement module(DEM)is designed by a convolutional layer to supplement frequency-domain information as well as providing nonlinear mapping that is beneficial for AMC.Multimodal network is designed to have multiple residual blocks,where each residual block has multiple convolutional kernels of different sizes for diverse feature extraction.Moreover,a deep supervised loss function is designed to supervise all parts of the network including the hidden layers and the DEM.Since different model may output different results,cooperative classifier is designed to avoid the randomness of single model and improve the reliability.Simulation results show that this few-shot learning-based AMC method can significantly improve the AMC accuracy compared to the existing methods.展开更多
The movements and intentions of other ships can be determined by gathering and examining ship sound signals.The extraction and analysis of ship sound signals fundamentally support the autonomous navigation of intellig...The movements and intentions of other ships can be determined by gathering and examining ship sound signals.The extraction and analysis of ship sound signals fundamentally support the autonomous navigation of intelligent ships.Mel scale frequency cepstral coefficient(MFCC)feature parameters are improved and optimized to form NewMFCC by introducing second-order difference and wavelet packet decomposition transformation methods in this paper.Transforming sound signals into a feature vector that fully describes the dynamic characteristics of ship sound signals and the high-and low-frequency information solves the problem of the inability to transport ordinary sound signals directly as signals for training in machine learning models.Radial basis function kernels are used to conduct support vector machine classifier simulation experiments.Five types of sound signals,namely,one type of ship sound signals and four types of interference sound signals,are categorized and identified as classification targets to verify the feasibility of the classification of ship sound signals and interference signals.The proposed method improves classification accuracy by approximately 15%.展开更多
With the rapid development of social economy,the society has entered into a new stage of development,especially in new media under the background of rapid development,makes the importance of news and information to ge...With the rapid development of social economy,the society has entered into a new stage of development,especially in new media under the background of rapid development,makes the importance of news and information to get the comprehensive promotion,and in order to further identify the positive and negative news,should be fully using machine learning methods,based on the emotion to realize the automatic classifying of news,in order to improve the efficiency of news classification.Therefore,the article first makes clear the basic outline of news sentiment classification.Secondly,the specific way of automatic classification of news emotion is deeply analyzed.On the basis of this,the paper puts forward the concrete measures of automatic classification of news emotion by using machine learning.展开更多
An automatic method for classifying frequency shift keying(FSK),minimum shift keying(MSK),phase shift keying(PSK),quadrature amplitude modulation(QAM),and orthogonal frequency division multiplexing(OFDM)is proposed by...An automatic method for classifying frequency shift keying(FSK),minimum shift keying(MSK),phase shift keying(PSK),quadrature amplitude modulation(QAM),and orthogonal frequency division multiplexing(OFDM)is proposed by simultaneously using normality test,spectral analysis,and geometrical characteristics of in-phase-quadrature(I-Q)constellation diagram.Since the extracted features are unique for each modulation,they can be considered as a fingerprint of each modulation.We show that the proposed algorithm outperforms the previously published methods in terms of signal-to-noise ratio(SNR)and success rate.For example,the success rate of the proposed method for 64-QAM modulation at SNR=11 dB is 99%.Another advantage of the proposed method is its wide SNR range;such that the probability of classification for 16-QAM at SNR=3 dB is almost 1.The proposed method also provides a database for geometrical features of I-Q constellation diagram.By comparing and correlating the data of the provided database with the estimated I-Q diagram of the received signal,the processing gain of 4 dB is obtained.Whatever can be mentioned about the preference of the proposed algorithm are low complexity,low SNR,wide range of modulation set,and enhanced recognition at higher-order modulations.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
Automatic modulation classification is the process of identification of the modulation type of a signal in a general environment. This paper proposes a new method to evaluate the tracking performance of large margin c...Automatic modulation classification is the process of identification of the modulation type of a signal in a general environment. This paper proposes a new method to evaluate the tracking performance of large margin classifier against signal-tonoise ratio （SNR）, and classifies all forms of primary user＇s signals in a cognitive radio environment. For achieving this objective, two structures of a large margin are developed in additive white Gaussian noise （AWGN） channels with priori unknown SNR. A combination of higher order statistics and instantaneous characteristics is selected as effective features. Simulation results show that the classification rates of the proposed structures are well robust against environmental SNR changes.展开更多
The detection of stellar flares is crucial to understanding dynamic processes at the stellar surface and their potential impact on surrounding exoplanetary systems.Extensive time series data acquired by the Transiting...The detection of stellar flares is crucial to understanding dynamic processes at the stellar surface and their potential impact on surrounding exoplanetary systems.Extensive time series data acquired by the Transiting Exoplanet Survey Satellite(TESS)offer valuable opportunities for large-scale flare studies.A variety of methods is currently employed for flare detection,with machine learning(ML)approaches demonstrating strong potential for automated classification tasks,particularly for the analysis of astronomical time series.This review provides an overview of the methods used to detect stellar flares in TESS data and evaluates their performance and effectiveness.It includes our assessment of both traditional detection techniques and more recent methods,such as ML algorithms,highlighting their strengths and limitations.By addressing current challenges and identifying promising approaches,this manuscript aims to support further studies and promote the development of stellar flare research.展开更多
Developing approaches to automate the analysis of the massive amounts of data sent back from the Moon will generate significant benefits for the field of lunar geomorphology. In this paper, we outline an automated met...Developing approaches to automate the analysis of the massive amounts of data sent back from the Moon will generate significant benefits for the field of lunar geomorphology. In this paper, we outline an automated method for mapping lunar landforms that is based on digital terrain analysis. An iterative self-organizing （ISO） cluster unsupervised classification enables the automatic mapping of landforms via a series of input raster bands that utilize six geomorphometric parameters. These parameters divide landforms into a number of spatially extended, topographically homogeneous segments that exhibit similar terrain attributes and neighborhood properties. To illustrate the applicability of our approach, we apply it to three representative test sites on the Moon, automatically presenting our results as a thematic landform map. We also quantitatively evaluated this approach using a series of confusion matrices, achieving overall accuracies as high as 83.34% and Kappa coefficients （K） as high as 0.77. An immediate version of our algorithm can also be applied for automatically mapping large-scale lunar landforms and for the quantitative comparison of lunar surface morphologies.展开更多
Automatic modulation classification(AMC)serves a challenging yet crucial role in wireless communications.Despite deep learning-based approaches being widely used in signal processing,they are challenged by signal dist...Automatic modulation classification(AMC)serves a challenging yet crucial role in wireless communications.Despite deep learning-based approaches being widely used in signal processing,they are challenged by signal distribution variations,especially in various channel conditions.In this paper,we introduce an adversarial transfer framework named frequency-learning adversarial networks(FLANs)based on transfer learning for cross-scenario signal classification.This method uses the stability in the frequency spectrum by introducing a frequency adaptation(FA)technique to incorporate target channel information into source-domain signals.To address the unpredictable interference in the channel,a fitting channel adaptation(FCA)module is used to reduce the difference between the source and target domains caused by variations in the channel environment.Experimental results illustrate that FLANs outperforms state-of-the-art transfer approaches,demonstrating an improved top-1 classification accuracy by about 5.2 percentage points in high signal-to-noise ratio(SNR)scenes on a cross-scenario real collected dataset CSRC2023.展开更多
With the rapid development of the Internet, text information has shown a blowout growth. Massivetext data such as news, social media posts, academic literature, etc. are constantly emerging, and manual classificationa...With the rapid development of the Internet, text information has shown a blowout growth. Massivetext data such as news, social media posts, academic literature, etc. are constantly emerging, and manual classificationand management of these texts has become time-consuming and inefficient, which is difficult to meetthe actual needs. The continuous progress of natural language processing technology, especially the rise of deeplearning methods, provides strong technical support for automatic text classification. Deep learning models canautomatically mine the essential features of text from massive samples, capture deep semantic representationinformation, and avoid the tedious process of manual design rules and features. In practical applications, textdata often co-exists with data of other modes (such as images, audio, etc.). Through the feature learning ofmultimodal data, the information of multiple modes can be mapped to the joint vector space, and the unifiedrepresentation of data can be obtained, so that the text classification can be more accurate. In recent years,pre-trained language models such as BERT and GPT have achieved remarkable results. These models learn acommon language representation through unsupervised pre-training on large-scale corpus, and then fine-tuneon specific text classification tasks, which can significantly improve the classification performance and furtherpromote the research of automatic text classification. Automatic text classification can classify massive textdata into different categories quickly and accurately, which is convenient for information storage, retrievaland management. For example, in the fields of library document management and enterprise document management,automatic classification can greatly improve work efficiency and save labor costs. In social mediaand online public opinion monitoring, automatic text classification can quickly identify text information withdifferent themes and emotional tendencies. This helps to timely understand the dynamics of public opinion,and provides a basis for the government, enterprises and other institutions to formulate corresponding copingstrategies. In the field of customer service, such as online customer service, customer feedback processing,etc., automatic text classification can automatically identify the types of questions and emotional tendencies ofcustomers. Thus, automated customer consultation and problem classification can be realized to improve theefficiency and quality of customer service. Automatic text classification is an important task in the field of naturallanguage processing, and its research progress can provide reference for other natural language processingtasks. For example, in tasks such as sentiment analysis, machine translation, question answering system, etc.,the techniques and methods of text classification can be applied and expanded. Automatic text classificationtechnology can be widely used in many fields, such as financial risk assessment, medical text analysis, legaldocument classification and so on. In these fields, automatic text classification can help professionals quicklysift and process a large amount of text information, improve work efficiency and decision-making accuracy.展开更多
An intelligent lithology identification method is proposed based on deep learning of the rock microscopic images.Based on the characteristics of rock images in the dataset,we used Xception,MobileNet_v2,Inception_ResNe...An intelligent lithology identification method is proposed based on deep learning of the rock microscopic images.Based on the characteristics of rock images in the dataset,we used Xception,MobileNet_v2,Inception_ResNet_v2,Inception_v3,Densenet121,ResNet101_v2,and ResNet-101 to develop microscopic image classification models,and then the network structures of seven different convolutional neural networks(CNNs)were compared.It shows that the multi-layer representation of rock features can be represented through convolution structures,thus better feature robustness can be achieved.For the loss function,cross-entropy is used to back propagate the weight parameters layer by layer,and the accuracy of the network is improved by frequent iterative training.We expanded a self-built dataset by using transfer learning and data augmentation.Next,accuracy(acc)and frames per second(fps)were used as the evaluation indexes to assess the accuracy and speed of model identification.The results show that the Xception-based model has the optimum performance,with an accuracy of 97.66%in the training dataset and 98.65%in the testing dataset.Furthermore,the fps of the model is 50.76,and the model is feasible to deploy under different hardware conditions and meets the requirements of rapid lithology identification.This proposed method is proved to be robust and versatile in generalization performance,and it is suitable for both geologists and engineers to identify lithology quickly.展开更多
The paper presents the results of a study of the female urethra in cases of urethral pain syndrome(UPS)and inflammatory diseases of the lower urinary tract using cross-polarization optical coherence tomography(CP OCT)...The paper presents the results of a study of the female urethra in cases of urethral pain syndrome(UPS)and inflammatory diseases of the lower urinary tract using cross-polarization optical coherence tomography(CP OCT).Urethral wall structure was studied in 86 patients;233 CP OCT images were collected.A comparative qualitative analysis of three groups of CP OCT images—"norm","Inflammation"and"UPS"—identified that despite the absence of a clear inflammatory factor in the patient's examination,the urethral tissues in UPS were in an altered state.The changes in the urethral wall with UPS and in cases of inflammation were similar.Using a point scale,three experts independently performed visual scoring of the CP OCT images.Three parameters:epithelial contrast,cavities and the minimum signal depth in the co-channel were evaluated.It was found that,individually,the parameters correlate only weakly with the diag-nosis.Area under the receiver operating characteristic(ROC)curve was from 0.51 to 0.78.The joint use of a number of visual signs has a greater diagnostic value than the use of the criteria individually.Area under the ROC curve using the developed cumulative criterion reached up to 0.87-0.89.This study could form the basis of a scoring system for assessing the state of the urethral tract using CP OCT images in real time.The CP OCT method provides information on the state of urethral tissues that cannot be obtained with traditional cystoscopy.展开更多
A low-complexity likelihood methodology is proposed for automatic modulation classification of orthogonal space-time block code(STBC)based multiple-input multiple-output(MIMO)systems.We exploit the zero-forcing equali...A low-complexity likelihood methodology is proposed for automatic modulation classification of orthogonal space-time block code(STBC)based multiple-input multiple-output(MIMO)systems.We exploit the zero-forcing equalization technique to modify the typical average likelihood ratio test(ALRT)function.The proposed ALRT function has a low computational complexity compared to existing ALRT functions for MIMO systems classification.The proposed approach is analyzed for blind channel scenarios when the receiver has imperfect channel state information(CSI).Performance analysis is carried out for scenarios with different numbers of antennas.Alamouti-STBC systems with 2×2 and 2×1 and space-time transmit diversity with a 4×4 transmit and receive antenna configuration are considered to verify the proposed approach.Some popular modulation schemes are used as the modulation test pool.Monte-Carlo simulations are performed to evaluate the proposed methodology,using the probability of correct classification as the criterion.Simulation results show that the proposed approach has high classification accuracy at low signal-to-noise ratios and exhibits robust behavior against high CSI estimation error variance.展开更多
Deep learning(DL) requires massive volume of data to train the network. Insufficient training data will cause serious overfitting problem and degrade the classification accuracy. In order to solve this problem, a meth...Deep learning(DL) requires massive volume of data to train the network. Insufficient training data will cause serious overfitting problem and degrade the classification accuracy. In order to solve this problem, a method for automatic modulation classification(AMC) using AlexNet with data augmentation was proposed. Three data augmentation methods is considered, i.e., random erasing, CutMix, and rotation. Firstly, modulated signals are converted into constellation representations. And all constellation representations are divided into training dataset and test dataset. Then, training dataset are augmented by three methods. Secondly, the optimal value of execution probability for random erasing and CutMix are determined. Simulation results show that both of them perform optimally when execution probability is 0.5. Thirdly, the performance of three data augmentation methods are evaluated. Simulation results demonstrate that all augmentation methods can improve the classification accuracy. Rotation improves the classification accuracy by 13.04% when signal noise ratio(SNR) is 2 dB. Among three methods, rotation outperforms random erasing and CutMix when SNR is greater than-6 dB. Finally, compared with other classification algorithms, random erasing, CutMix, and rotation used in this paper achieved the performance significantly improved. It is worth mentioning that the classification accuracy can reach 90.5% with SNR at 10 dB.展开更多
The Sentinel-2 satellites are providing an unparalleled wealth of high-resolution remotely sensed information with a short revisit cycle, which is ideal for mapping burned areas both accurately and timely. However, th...The Sentinel-2 satellites are providing an unparalleled wealth of high-resolution remotely sensed information with a short revisit cycle, which is ideal for mapping burned areas both accurately and timely. However, the high detail and volume of information provided actually encumbers the automation of the mapping process, at least for the level of automation required to map systematically wildfires on a national level. This paper proposes a fully automated methodology for mapping burn scars using Sentinel-2 data. Information extracted from a pair of Sentinel-2 images, one pre-fire and one post-fire, is jointly used to automatically label a set of training patterns via two empirical rules. An initial pixel-based classification is derived using this training set by means of a Support Vector Machine (SVM) classifier. The latter is subsequently smoothed following a multiple spectral-spatial classification (MSSC) approach, which increases the mapping accuracy and thematic consistency of the final burned area delineation. The proposed methodology was tested on six recent wildfire events in Greece, selected to cover representative cases of the Greek ecosystems and to present challenges in burned area mapping. The lowest classification accuracy achieved was 92%, whereas Matthews correlation coefficient (MCC) was greater or equal to 0.85.展开更多
Defects detection with Electroluminescence(EL)image for photovoltaic(PV)module has become a standard test procedure during the process of production,installation,and operation of solar modules.There are some typical d...Defects detection with Electroluminescence(EL)image for photovoltaic(PV)module has become a standard test procedure during the process of production,installation,and operation of solar modules.There are some typical defects types,such as crack,finger interruption,that can be recognized with high accuracy.However,due to the complexity of EL images and the limitation of the dataset,it is hard to label all types of defects during the inspection process.The unknown or unlabeled create significant difficulties in the practical application of the automatic defects detection technique.To address the problem,we proposed an evolutionary algorithm combined with traditional image processing technology,deep learning,transfer learning,and deep clustering,which can recognize the unknown or unlabeled in the original dataset defects automatically along with the increasing of the dataset size.Specifically,we first propose a deep learning-based features extractor and defects classifier.Then,the unlabeled defects can be classified by the deep clustering algorithm and stored separately to update the original database without human intervention.When the number of unknown images reaches the preset values,transfer learning is introduced to train the classifier with the updated database.The fine-tuned model can detect new defects with high accuracy.Finally,numerical results confirm that the proposed solution can carry out efficient and accurate defect detection automatically using electroluminescence images.展开更多
Although, researchers in the ATC field have done a wide range of work based on SVM, almost all existing approaches utilize an empirical model of selection algorithms. Their attempts to model automatic selection in pra...Although, researchers in the ATC field have done a wide range of work based on SVM, almost all existing approaches utilize an empirical model of selection algorithms. Their attempts to model automatic selection in practical, large-scale, text classification systems have been limited. In this paper, we propose a new model selection algorithm that utilizes the DDAG learning architecture. This architecture derives a new large-scale text classifier with very good performance. Experimental results show that the proposed algorithm has good efficiency and the necessary generalization capability while handling large-scale multi-class text classification tasks.展开更多
文摘Purpose:With more and more digital collections of various information resources becoming available,also increasing is the challenge of assigning subject index terms and classes from quality knowledge organization systems.While the ultimate purpose is to understand the value of automatically produced Dewey Decimal Classification(DDC)classes for Swedish digital collections,the paper aims to evaluate the performance of six machine learning algorithms as well as a string-matching algorithm based on characteristics of DDC.Design/methodology/approach:State-of-the-art machine learning algorithms require at least 1,000 training examples per class.The complete data set at the time of research involved 143,838 records which had to be reduced to top three hierarchical levels of DDC in order to provide sufficient training data(totaling 802 classes in the training and testing sample,out of 14,413 classes at all levels).Findings:Evaluation shows that Support Vector Machine with linear kernel outperforms other machine learning algorithms as well as the string-matching algorithm on average;the string-matching algorithm outperforms machine learning for specific classes when characteristics of DDC are most suitable for the task.Word embeddings combined with different types of neural networks(simple linear network,standard neural network,1 D convolutional neural network,and recurrent neural network)produced worse results than Support Vector Machine,but reach close results,with the benefit of a smaller representation size.Impact of features in machine learning shows that using keywords or combining titles and keywords gives better results than using only titles as input.Stemming only marginally improves the results.Removed stop-words reduced accuracy in most cases,while removing less frequent words increased it marginally.The greatest impact is produced by the number of training examples:81.90%accuracy on the training set is achieved when at least 1,000 records per class are available in the training set,and 66.13%when too few records(often less than A Comparison of Approaches100 per class)on which to train are available—and these hold only for top 3 hierarchical levels(803 instead of 14,413 classes).Research limitations:Having to reduce the number of hierarchical levels to top three levels of DDC because of the lack of training data for all classes,skews the results so that they work in experimental conditions but barely for end users in operational retrieval systems.Practical implications:In conclusion,for operative information retrieval systems applying purely automatic DDC does not work,either using machine learning(because of the lack of training data for the large number of DDC classes)or using string-matching algorithm(because DDC characteristics perform well for automatic classification only in a small number of classes).Over time,more training examples may become available,and DDC may be enriched with synonyms in order to enhance accuracy of automatic classification which may also benefit information retrieval performance based on DDC.In order for quality information services to reach the objective of highest possible precision and recall,automatic classification should never be implemented on its own;instead,machine-aided indexing that combines the efficiency of automatic suggestions with quality of human decisions at the final stage should be the way for the future.Originality/value:The study explored machine learning on a large classification system of over 14,000 classes which is used in operational information retrieval systems.Due to lack of sufficient training data across the entire set of classes,an approach complementing machine learning,that of string matching,was applied.This combination should be explored further since it provides the potential for real-life applications with large target classification systems.
文摘To promote reliable and secure communications in the cognitive radio network,the automatic modulation classification algorithms have been mainly proposed to estimate a single modulation.In this paper,we address the classification of superimposed modulations dedicated to 5G multipleinput multiple-output(MIMO)two-way cognitive relay network in realistic channels modeled with Nakagami-m distribution.Our purpose consists of classifying pairs of users modulations from superimposed signals.To achieve this goal,we apply the higher-order statistics in conjunction with the Multi-BoostAB classifier.We use several efficiency metrics including the true positive(TP)rate,false positive(FP)rate,precision,recall,F-Measure and receiver operating characteristic(ROC)area in order to evaluate the performance of the proposed algorithm in terms of correct superimposed modulations classification.Computer simulations prove that our proposal allows obtaining a good probability of classification for ten superimposed modulations at a low signal-to-noise ratio,including the worst case(i.e.,m=0.5),where the fading distribution follows a one-sided Gaussian distribution.We also carry out a comparative study between our proposal usingMultiBoostAB classifier with the decision tree(J48)classifier.Simulation results show that the performance of MultiBoostAB on the superimposed modulations classifications outperforms the one of J48 classifier.In addition,we study the impact of the symbols number,path loss exponent and relay position on the performance of the proposed automatic classification superimposed modulations in terms of probability of correct classification.
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 in part by National Key Research and Development Program of China under Grant 2021YFB2900404.
文摘Due to the limited computational capability and the diversity of the Internet of Things devices working in different environment,we consider fewshot learning-based automatic modulation classification(AMC)to improve its reliability.A data enhancement module(DEM)is designed by a convolutional layer to supplement frequency-domain information as well as providing nonlinear mapping that is beneficial for AMC.Multimodal network is designed to have multiple residual blocks,where each residual block has multiple convolutional kernels of different sizes for diverse feature extraction.Moreover,a deep supervised loss function is designed to supervise all parts of the network including the hidden layers and the DEM.Since different model may output different results,cooperative classifier is designed to avoid the randomness of single model and improve the reliability.Simulation results show that this few-shot learning-based AMC method can significantly improve the AMC accuracy compared to the existing methods.
文摘The movements and intentions of other ships can be determined by gathering and examining ship sound signals.The extraction and analysis of ship sound signals fundamentally support the autonomous navigation of intelligent ships.Mel scale frequency cepstral coefficient(MFCC)feature parameters are improved and optimized to form NewMFCC by introducing second-order difference and wavelet packet decomposition transformation methods in this paper.Transforming sound signals into a feature vector that fully describes the dynamic characteristics of ship sound signals and the high-and low-frequency information solves the problem of the inability to transport ordinary sound signals directly as signals for training in machine learning models.Radial basis function kernels are used to conduct support vector machine classifier simulation experiments.Five types of sound signals,namely,one type of ship sound signals and four types of interference sound signals,are categorized and identified as classification targets to verify the feasibility of the classification of ship sound signals and interference signals.The proposed method improves classification accuracy by approximately 15%.
文摘With the rapid development of social economy,the society has entered into a new stage of development,especially in new media under the background of rapid development,makes the importance of news and information to get the comprehensive promotion,and in order to further identify the positive and negative news,should be fully using machine learning methods,based on the emotion to realize the automatic classifying of news,in order to improve the efficiency of news classification.Therefore,the article first makes clear the basic outline of news sentiment classification.Secondly,the specific way of automatic classification of news emotion is deeply analyzed.On the basis of this,the paper puts forward the concrete measures of automatic classification of news emotion by using machine learning.
文摘An automatic method for classifying frequency shift keying(FSK),minimum shift keying(MSK),phase shift keying(PSK),quadrature amplitude modulation(QAM),and orthogonal frequency division multiplexing(OFDM)is proposed by simultaneously using normality test,spectral analysis,and geometrical characteristics of in-phase-quadrature(I-Q)constellation diagram.Since the extracted features are unique for each modulation,they can be considered as a fingerprint of each modulation.We show that the proposed algorithm outperforms the previously published methods in terms of signal-to-noise ratio(SNR)and success rate.For example,the success rate of the proposed method for 64-QAM modulation at SNR=11 dB is 99%.Another advantage of the proposed method is its wide SNR range;such that the probability of classification for 16-QAM at SNR=3 dB is almost 1.The proposed method also provides a database for geometrical features of I-Q constellation diagram.By comparing and correlating the data of the provided database with the estimated I-Q diagram of the received signal,the processing gain of 4 dB is obtained.Whatever can be mentioned about the preference of the proposed algorithm are low complexity,low SNR,wide range of modulation set,and enhanced recognition at higher-order modulations.
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.
文摘Automatic modulation classification is the process of identification of the modulation type of a signal in a general environment. This paper proposes a new method to evaluate the tracking performance of large margin classifier against signal-tonoise ratio （SNR）, and classifies all forms of primary user＇s signals in a cognitive radio environment. For achieving this objective, two structures of a large margin are developed in additive white Gaussian noise （AWGN） channels with priori unknown SNR. A combination of higher order statistics and instantaneous characteristics is selected as effective features. Simulation results show that the classification rates of the proposed structures are well robust against environmental SNR changes.
基金supported by the National Natural Science Foundation of China(12473104 and U2031144).
文摘The detection of stellar flares is crucial to understanding dynamic processes at the stellar surface and their potential impact on surrounding exoplanetary systems.Extensive time series data acquired by the Transiting Exoplanet Survey Satellite(TESS)offer valuable opportunities for large-scale flare studies.A variety of methods is currently employed for flare detection,with machine learning(ML)approaches demonstrating strong potential for automated classification tasks,particularly for the analysis of astronomical time series.This review provides an overview of the methods used to detect stellar flares in TESS data and evaluates their performance and effectiveness.It includes our assessment of both traditional detection techniques and more recent methods,such as ML algorithms,highlighting their strengths and limitations.By addressing current challenges and identifying promising approaches,this manuscript aims to support further studies and promote the development of stellar flare research.
基金Foundation： National Natural Science Foundation of China, No.41571388 National Special Basic Research Fund, No.2015FY210500
文摘Developing approaches to automate the analysis of the massive amounts of data sent back from the Moon will generate significant benefits for the field of lunar geomorphology. In this paper, we outline an automated method for mapping lunar landforms that is based on digital terrain analysis. An iterative self-organizing （ISO） cluster unsupervised classification enables the automatic mapping of landforms via a series of input raster bands that utilize six geomorphometric parameters. These parameters divide landforms into a number of spatially extended, topographically homogeneous segments that exhibit similar terrain attributes and neighborhood properties. To illustrate the applicability of our approach, we apply it to three representative test sites on the Moon, automatically presenting our results as a thematic landform map. We also quantitatively evaluated this approach using a series of confusion matrices, achieving overall accuracies as high as 83.34% and Kappa coefficients （K） as high as 0.77. An immediate version of our algorithm can also be applied for automatically mapping large-scale lunar landforms and for the quantitative comparison of lunar surface morphologies.
基金Project supported by the National Natural Science Foundation of China(No.U20B2042)。
文摘Automatic modulation classification(AMC)serves a challenging yet crucial role in wireless communications.Despite deep learning-based approaches being widely used in signal processing,they are challenged by signal distribution variations,especially in various channel conditions.In this paper,we introduce an adversarial transfer framework named frequency-learning adversarial networks(FLANs)based on transfer learning for cross-scenario signal classification.This method uses the stability in the frequency spectrum by introducing a frequency adaptation(FA)technique to incorporate target channel information into source-domain signals.To address the unpredictable interference in the channel,a fitting channel adaptation(FCA)module is used to reduce the difference between the source and target domains caused by variations in the channel environment.Experimental results illustrate that FLANs outperforms state-of-the-art transfer approaches,demonstrating an improved top-1 classification accuracy by about 5.2 percentage points in high signal-to-noise ratio(SNR)scenes on a cross-scenario real collected dataset CSRC2023.
文摘With the rapid development of the Internet, text information has shown a blowout growth. Massivetext data such as news, social media posts, academic literature, etc. are constantly emerging, and manual classificationand management of these texts has become time-consuming and inefficient, which is difficult to meetthe actual needs. The continuous progress of natural language processing technology, especially the rise of deeplearning methods, provides strong technical support for automatic text classification. Deep learning models canautomatically mine the essential features of text from massive samples, capture deep semantic representationinformation, and avoid the tedious process of manual design rules and features. In practical applications, textdata often co-exists with data of other modes (such as images, audio, etc.). Through the feature learning ofmultimodal data, the information of multiple modes can be mapped to the joint vector space, and the unifiedrepresentation of data can be obtained, so that the text classification can be more accurate. In recent years,pre-trained language models such as BERT and GPT have achieved remarkable results. These models learn acommon language representation through unsupervised pre-training on large-scale corpus, and then fine-tuneon specific text classification tasks, which can significantly improve the classification performance and furtherpromote the research of automatic text classification. Automatic text classification can classify massive textdata into different categories quickly and accurately, which is convenient for information storage, retrievaland management. For example, in the fields of library document management and enterprise document management,automatic classification can greatly improve work efficiency and save labor costs. In social mediaand online public opinion monitoring, automatic text classification can quickly identify text information withdifferent themes and emotional tendencies. This helps to timely understand the dynamics of public opinion,and provides a basis for the government, enterprises and other institutions to formulate corresponding copingstrategies. In the field of customer service, such as online customer service, customer feedback processing,etc., automatic text classification can automatically identify the types of questions and emotional tendencies ofcustomers. Thus, automated customer consultation and problem classification can be realized to improve theefficiency and quality of customer service. Automatic text classification is an important task in the field of naturallanguage processing, and its research progress can provide reference for other natural language processingtasks. For example, in tasks such as sentiment analysis, machine translation, question answering system, etc.,the techniques and methods of text classification can be applied and expanded. Automatic text classificationtechnology can be widely used in many fields, such as financial risk assessment, medical text analysis, legaldocument classification and so on. In these fields, automatic text classification can help professionals quicklysift and process a large amount of text information, improve work efficiency and decision-making accuracy.
基金support from the National Natural Science Foundation of China(Grant Nos.52022053 and 52009073)the Natural Science Foundation of Shandong Province(Grant No.ZR201910270116).
文摘An intelligent lithology identification method is proposed based on deep learning of the rock microscopic images.Based on the characteristics of rock images in the dataset,we used Xception,MobileNet_v2,Inception_ResNet_v2,Inception_v3,Densenet121,ResNet101_v2,and ResNet-101 to develop microscopic image classification models,and then the network structures of seven different convolutional neural networks(CNNs)were compared.It shows that the multi-layer representation of rock features can be represented through convolution structures,thus better feature robustness can be achieved.For the loss function,cross-entropy is used to back propagate the weight parameters layer by layer,and the accuracy of the network is improved by frequent iterative training.We expanded a self-built dataset by using transfer learning and data augmentation.Next,accuracy(acc)and frames per second(fps)were used as the evaluation indexes to assess the accuracy and speed of model identification.The results show that the Xception-based model has the optimum performance,with an accuracy of 97.66%in the training dataset and 98.65%in the testing dataset.Furthermore,the fps of the model is 50.76,and the model is feasible to deploy under different hardware conditions and meets the requirements of rapid lithology identification.This proposed method is proved to be robust and versatile in generalization performance,and it is suitable for both geologists and engineers to identify lithology quickly.
基金financially supported by the the R.FBR grant,project No.19-07-00395
文摘The paper presents the results of a study of the female urethra in cases of urethral pain syndrome(UPS)and inflammatory diseases of the lower urinary tract using cross-polarization optical coherence tomography(CP OCT).Urethral wall structure was studied in 86 patients;233 CP OCT images were collected.A comparative qualitative analysis of three groups of CP OCT images—"norm","Inflammation"and"UPS"—identified that despite the absence of a clear inflammatory factor in the patient's examination,the urethral tissues in UPS were in an altered state.The changes in the urethral wall with UPS and in cases of inflammation were similar.Using a point scale,three experts independently performed visual scoring of the CP OCT images.Three parameters:epithelial contrast,cavities and the minimum signal depth in the co-channel were evaluated.It was found that,individually,the parameters correlate only weakly with the diag-nosis.Area under the receiver operating characteristic(ROC)curve was from 0.51 to 0.78.The joint use of a number of visual signs has a greater diagnostic value than the use of the criteria individually.Area under the ROC curve using the developed cumulative criterion reached up to 0.87-0.89.This study could form the basis of a scoring system for assessing the state of the urethral tract using CP OCT images in real time.The CP OCT method provides information on the state of urethral tissues that cannot be obtained with traditional cystoscopy.
基金Project supported by the National Natural Science Foundation of China(Nos.61172078,61571224,and 61571225)Six Talent Peaks Pro ject in Jiangsu Province,China.
文摘A low-complexity likelihood methodology is proposed for automatic modulation classification of orthogonal space-time block code(STBC)based multiple-input multiple-output(MIMO)systems.We exploit the zero-forcing equalization technique to modify the typical average likelihood ratio test(ALRT)function.The proposed ALRT function has a low computational complexity compared to existing ALRT functions for MIMO systems classification.The proposed approach is analyzed for blind channel scenarios when the receiver has imperfect channel state information(CSI).Performance analysis is carried out for scenarios with different numbers of antennas.Alamouti-STBC systems with 2×2 and 2×1 and space-time transmit diversity with a 4×4 transmit and receive antenna configuration are considered to verify the proposed approach.Some popular modulation schemes are used as the modulation test pool.Monte-Carlo simulations are performed to evaluate the proposed methodology,using the probability of correct classification as the criterion.Simulation results show that the proposed approach has high classification accuracy at low signal-to-noise ratios and exhibits robust behavior against high CSI estimation error variance.
基金supported by the National Key Research and Development Program of China (2019YFC1511300)the Science and Technology Research Program of Chongqing Municipal Education Commission (KJQN201800621)the Science Foundation of Chongqing University of Posts and Telecommunications (XJG20103)。
文摘Deep learning(DL) requires massive volume of data to train the network. Insufficient training data will cause serious overfitting problem and degrade the classification accuracy. In order to solve this problem, a method for automatic modulation classification(AMC) using AlexNet with data augmentation was proposed. Three data augmentation methods is considered, i.e., random erasing, CutMix, and rotation. Firstly, modulated signals are converted into constellation representations. And all constellation representations are divided into training dataset and test dataset. Then, training dataset are augmented by three methods. Secondly, the optimal value of execution probability for random erasing and CutMix are determined. Simulation results show that both of them perform optimally when execution probability is 0.5. Thirdly, the performance of three data augmentation methods are evaluated. Simulation results demonstrate that all augmentation methods can improve the classification accuracy. Rotation improves the classification accuracy by 13.04% when signal noise ratio(SNR) is 2 dB. Among three methods, rotation outperforms random erasing and CutMix when SNR is greater than-6 dB. Finally, compared with other classification algorithms, random erasing, CutMix, and rotation used in this paper achieved the performance significantly improved. It is worth mentioning that the classification accuracy can reach 90.5% with SNR at 10 dB.
文摘The Sentinel-2 satellites are providing an unparalleled wealth of high-resolution remotely sensed information with a short revisit cycle, which is ideal for mapping burned areas both accurately and timely. However, the high detail and volume of information provided actually encumbers the automation of the mapping process, at least for the level of automation required to map systematically wildfires on a national level. This paper proposes a fully automated methodology for mapping burn scars using Sentinel-2 data. Information extracted from a pair of Sentinel-2 images, one pre-fire and one post-fire, is jointly used to automatically label a set of training patterns via two empirical rules. An initial pixel-based classification is derived using this training set by means of a Support Vector Machine (SVM) classifier. The latter is subsequently smoothed following a multiple spectral-spatial classification (MSSC) approach, which increases the mapping accuracy and thematic consistency of the final burned area delineation. The proposed methodology was tested on six recent wildfire events in Greece, selected to cover representative cases of the Greek ecosystems and to present challenges in burned area mapping. The lowest classification accuracy achieved was 92%, whereas Matthews correlation coefficient (MCC) was greater or equal to 0.85.
文摘Defects detection with Electroluminescence(EL)image for photovoltaic(PV)module has become a standard test procedure during the process of production,installation,and operation of solar modules.There are some typical defects types,such as crack,finger interruption,that can be recognized with high accuracy.However,due to the complexity of EL images and the limitation of the dataset,it is hard to label all types of defects during the inspection process.The unknown or unlabeled create significant difficulties in the practical application of the automatic defects detection technique.To address the problem,we proposed an evolutionary algorithm combined with traditional image processing technology,deep learning,transfer learning,and deep clustering,which can recognize the unknown or unlabeled in the original dataset defects automatically along with the increasing of the dataset size.Specifically,we first propose a deep learning-based features extractor and defects classifier.Then,the unlabeled defects can be classified by the deep clustering algorithm and stored separately to update the original database without human intervention.When the number of unknown images reaches the preset values,transfer learning is introduced to train the classifier with the updated database.The fine-tuned model can detect new defects with high accuracy.Finally,numerical results confirm that the proposed solution can carry out efficient and accurate defect detection automatically using electroluminescence images.
文摘Although, researchers in the ATC field have done a wide range of work based on SVM, almost all existing approaches utilize an empirical model of selection algorithms. Their attempts to model automatic selection in practical, large-scale, text classification systems have been limited. In this paper, we propose a new model selection algorithm that utilizes the DDAG learning architecture. This architecture derives a new large-scale text classifier with very good performance. Experimental results show that the proposed algorithm has good efficiency and the necessary generalization capability while handling large-scale multi-class text classification tasks.
