Vibration-based pavement condition(roughness and obvious anomalies)monitoring has been expanding in road engineering.However,the indistinctive transverse cracking has hardly been considered.Therefore,a vehicle-based n...Vibration-based pavement condition(roughness and obvious anomalies)monitoring has been expanding in road engineering.However,the indistinctive transverse cracking has hardly been considered.Therefore,a vehicle-based novel method is proposed for detecting the transverse cracking through signal processing techniques and support vector machine(SVM).The vibration signals of the car traveling on the transverse-cracked and the crack-free sections were subjected to signal processing in time domain,frequency domain and wavelet domain,aiming to find indices that can discriminate vibration signal between the cracked and uncracked section.These indices were used to form 8 SVM models.The model with the highest accuracy and F1-measure was preferred,consisting of features including vehicle speed,range,relative standard deviation,maximum Fourier coefficient,and wavelet coefficient.Therefore,a crack and crack-free classifier was developed.Then its feasibility was investigated by 2292 pavement sections.The detection accuracy and F1-measure are 97.25%and 85.25%,respectively.The cracking detection approach proposed in this paper and the smartphone-based detection method for IRI and other distress may form a comprehensive pavement condition survey system.展开更多
Moving object extraction and classification are important problems in automated video surveillance systems. A background model based on region segmentation is proposed. An adaptive single Gaussian background model is ...Moving object extraction and classification are important problems in automated video surveillance systems. A background model based on region segmentation is proposed. An adaptive single Gaussian background model is used in the stable region with gradual changes, and a nonparametric model is used in the variable region with jumping changes. A generalized agglomerative scheme is used to merge the pixels in the variable region and fill in the small interspaces. A two-threshold sequential algorithmic scheme is used to group the background samples of the variable region into distinct Gaussian distributions to accelerate the kernel density computation speed of the nonparametric model. In the feature-based object classification phase, the surveillance scene is first partitioned according to the road boundaries of different traffic directions and then re-segmented according to their scene localities. The method improves the discriminability of the features in each partition. AdaBoost method is applied to evaluate the relative importance of the features in each partition respectively and distinguish whether an object is a vehicle, a single human, a human group, or a bike. Experimental results show that the proposed method achieves higher performance in comparison with the existing method.展开更多
The Gaussian mixture model (GMM), k-nearest neighbor (k-NN), quadratic discriminant analysis (QDA), and linear discriminant analysis (LDA) were compared to classify wrist motions using surface electromyogram (EMG). Ef...The Gaussian mixture model (GMM), k-nearest neighbor (k-NN), quadratic discriminant analysis (QDA), and linear discriminant analysis (LDA) were compared to classify wrist motions using surface electromyogram (EMG). Effect of feature selection in EMG signal processing was also verified by comparing classification accuracy of each feature, and the enhancement of classification accuracy by normalization was confirmed. EMG signals were acquired from two electrodes placed on the forearm of twenty eight healthy subjects and used for recognition of wrist motion. Features were extracted from the obtained EMG signals in the time domain and were applied to classification methods. The difference absolute mean value (DAMV), difference absolute standard deviation value (DASDV), mean absolute value (MAV), root mean square (RMS) were used for composing 16 double features which were combined of two channels. In the classification methods, the highest accuracy of classification showed in the GMM. The most effective combination of classification method and double feature was (MAV, DAMV) of GMM and its classification accuracy was 96.85%. The results of normalization were better than those of non-normalization in GMM, k-NN, and LDA.展开更多
This article explores the ability of multivariate autoregressive model(MAR)and scalar AR model to extract the features from two-lead electrocardiogram signals in order to classify certain cardiac arrhythmias.The class...This article explores the ability of multivariate autoregressive model(MAR)and scalar AR model to extract the features from two-lead electrocardiogram signals in order to classify certain cardiac arrhythmias.The classification performance of four different ECG feature sets based on the model coefficients are shown.The data in the analysis including normal sinus rhythm, atria premature contraction,premature ventricular contraction,ventricular tachycardia,ventricular fibrillation and superventricular tachyeardia is obtained from the MIT-BIH database.The classification is performed using a quadratic diacriminant function.The results show the MAR coefficients produce the best results among the four ECG representations and the MAR modeling is a useful classification and diagnosis tool.展开更多
A model classification rate control method for video coding is proposed. The macro-blocks are classified according to their prediction errors, and different parameters are used in the rate-quantization and distortion-...A model classification rate control method for video coding is proposed. The macro-blocks are classified according to their prediction errors, and different parameters are used in the rate-quantization and distortion-quantization model. The different model parameters are calculated from the previous frame of the same type in the process of coding. These models are used to estimate the relations among rate, distortion and quantization of the current frame. Further steps, such as R-D optimization based quantization adjustment and smoothing of quantization of adjacent macrobloeks, are used to improve the quality. The results of the experiments prove that the technique is effective and can be realized easily. The method presented in the paper can be a good way for MPEG and H. 264 rate control.展开更多
The process of human natural scene categorization consists of two correlated stages: visual perception and visual cognition of natural scenes.Inspired by this fact,we propose a biologically plausible approach for natu...The process of human natural scene categorization consists of two correlated stages: visual perception and visual cognition of natural scenes.Inspired by this fact,we propose a biologically plausible approach for natural scene image classification.This approach consists of one visual perception model and two visual cognition models.The visual perception model,composed of two steps,is used to extract discriminative features from natural scene images.In the first step,we mimic the oriented and bandpass properties of human primary visual cortex by a special complex wavelets transform,which can decompose a natural scene image into a series of 2D spatial structure signals.In the second step,a hybrid statistical feature extraction method is used to generate gist features from those 2D spatial structure signals.Then we design a cognitive feedback model to realize adaptive optimization for the visual perception model.At last,we build a multiple semantics based cognition model to imitate human cognitive mode in rapid natural scene categorization.Experiments on natural scene datasets show that the proposed method achieves high efficiency and accuracy for natural scene classification.展开更多
[目的/意义]针对实体嵌套、实体类型混淆等问题导致的农业病害命名实体识别(Named Entities Recognition,NER)准确率不高的情况,以PointerNet为基准模型,提出一种基于RoFormer预训练模型的指针网络农业病害NER方法RoFormer-PointerNet。...[目的/意义]针对实体嵌套、实体类型混淆等问题导致的农业病害命名实体识别(Named Entities Recognition,NER)准确率不高的情况,以PointerNet为基准模型,提出一种基于RoFormer预训练模型的指针网络农业病害NER方法RoFormer-PointerNet。[方法]采用RoFormer预训练模型对输入的文本进行向量化,利用其独特的旋转位置嵌入方法来捕捉位置信息,丰富字词特征信息,从而解决一词多义导致的类型易混淆的问题。使用指针网络进行解码,利用指针网络的首尾指针标注方式抽取句子中的所有实体,首尾指针标注方式可以解决实体抽取中存在的嵌套问题。[结果和讨论]自建农业病害数据集,数据集中包含2867条标注语料,共10282个实体。为验证RoFormer预训练模型在实体抽取上的优越性,采用Word2Vec、BERT、RoBERTa等多种向量化模型进行对比试验,RoFormer-PointerNet与其他模型相比,模型精确率、召回率、F1值均为最优,分别为87.49%,85.76%和86.62%。为验证RoFormer-PointerNet在缓解实体嵌套的优势,与使用最为广泛的双向长短期记忆神经网络(Bidirectional Long Short-Term Memory,BiLSTM)和条件随机场(Conditional Random Field,CRF)模型进行对比试验,RoFormer-PointerNet比RoFormer-BiLSTM模型、RoFormer-CRF模型和RoFormer-BiLSTM-CRF模型分别高出4.8%、5.67%和3.87%,证明用指针网络模型可以很好解决实体嵌套问题。最后验证RoFormer-PointerNet方法在农业病害数据集中的识别性能,针对病害症状、病害名称、防治方法等8类实体进行了识别实验,本方法识别的精确率、召回率和F1值分别为87.49%、85.76%和86.62%,为同类最优。[结论]本研究提出的方法能有效识别中文农业病害文本中的实体,识别效果优于其他模型。在解决实体抽取过程中的实体嵌套和类型混淆等问题方面具有一定优势。展开更多
Influenza caused by influenza virus,seriously threaten human life and health.Drug treatment is one of the effective measurement.However,there are only two classes of drugs,one class is M2 blockers and another is neura...Influenza caused by influenza virus,seriously threaten human life and health.Drug treatment is one of the effective measurement.However,there are only two classes of drugs,one class is M2 blockers and another is neuraminidase(NA) inhibitors.The recent antiviral surveillance studies reported a global significant increase in M2 blocker resistance among influenza viruses,and the resistant virus strains against NA inhibitor are also reported in clinical treatment.Therefore thediscovery of new medicines with low resistance has become very urgent.As all known,traditional medicines with multi-target features and network mechanism often possess low resistance.Compound Yizhihao,which consists of radix isatidis,folium isatidis,Artemisia rupestris,is one of the famous traditional medicine for influenza treatment in China,however its mechanism of action against influenza is unclear.In this study,the multiple targets related with influenza disease and the known chemical constituents from Compound Yizhihao were collected,and multi-target QSAR(mt-QSAR) classification models were developed by Na?e Bayesian algorithm and verified by various datasets.Then the classification models were applied to predict the effective constituents and their drug targets.Finally,the constituent-target-pathway network was constructed,which revealed the effective constituents and their network mechanism in Compound Yizhihao.This study will lay important basis for the clinical uses for influenza treatment and for the further research and development of the effective constituents.展开更多
为解决医学文本使用预训练模型处理速度慢、对硬件的算力要求高、难以分辨少样本类别,以及使用传统小模型由于自身限制难以做到足够的准确度的问题,提出了一种融合预训练和元蒸馏的文本分类模型PTMD(fusion of pre-training and meta di...为解决医学文本使用预训练模型处理速度慢、对硬件的算力要求高、难以分辨少样本类别,以及使用传统小模型由于自身限制难以做到足够的准确度的问题,提出了一种融合预训练和元蒸馏的文本分类模型PTMD(fusion of pre-training and meta distillation model)。PTMD针对医学文本的多标签问题,通过对比训练对RoBERTa预训练方法进行微调,再由双向内置注意力简单循环单元充分获取语义信息。最后在传统蒸馏模型的基础上融合元学习和助教的思想,通过教学实验和双级模型等方法,提高模型的教学水平,最终在减少训练成本的基础上得到一个高性能医学文本分类模型。实验结果表明,教师模型在CHIP2019评测三数据集上的F1值达到了85.47%,同时学生模型在F1值损失1.45个百分点的情况下,将模型规模缩小到教师模型的近1/6,效果高于大多数传统预训练模型和知识蒸馏模型,证明了该模型具有良好的实用价值。展开更多
基金Project(51778482)supported by the National Natural Science Foundation of China。
文摘Vibration-based pavement condition(roughness and obvious anomalies)monitoring has been expanding in road engineering.However,the indistinctive transverse cracking has hardly been considered.Therefore,a vehicle-based novel method is proposed for detecting the transverse cracking through signal processing techniques and support vector machine(SVM).The vibration signals of the car traveling on the transverse-cracked and the crack-free sections were subjected to signal processing in time domain,frequency domain and wavelet domain,aiming to find indices that can discriminate vibration signal between the cracked and uncracked section.These indices were used to form 8 SVM models.The model with the highest accuracy and F1-measure was preferred,consisting of features including vehicle speed,range,relative standard deviation,maximum Fourier coefficient,and wavelet coefficient.Therefore,a crack and crack-free classifier was developed.Then its feasibility was investigated by 2292 pavement sections.The detection accuracy and F1-measure are 97.25%and 85.25%,respectively.The cracking detection approach proposed in this paper and the smartphone-based detection method for IRI and other distress may form a comprehensive pavement condition survey system.
基金supported by the Science and Technology Program of Zhejiang Province of China(2005C11001-02).
文摘Moving object extraction and classification are important problems in automated video surveillance systems. A background model based on region segmentation is proposed. An adaptive single Gaussian background model is used in the stable region with gradual changes, and a nonparametric model is used in the variable region with jumping changes. A generalized agglomerative scheme is used to merge the pixels in the variable region and fill in the small interspaces. A two-threshold sequential algorithmic scheme is used to group the background samples of the variable region into distinct Gaussian distributions to accelerate the kernel density computation speed of the nonparametric model. In the feature-based object classification phase, the surveillance scene is first partitioned according to the road boundaries of different traffic directions and then re-segmented according to their scene localities. The method improves the discriminability of the features in each partition. AdaBoost method is applied to evaluate the relative importance of the features in each partition respectively and distinguish whether an object is a vehicle, a single human, a human group, or a bike. Experimental results show that the proposed method achieves higher performance in comparison with the existing method.
基金Project(NIPA-2012-H0401-12-1007) supported by the MKE(The Ministry of Knowledge Economy), Korea, supervised by the NIPAProject(2010-0020163) supported by Key Research Institute Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education, Science and Technology, Korea
文摘The Gaussian mixture model (GMM), k-nearest neighbor (k-NN), quadratic discriminant analysis (QDA), and linear discriminant analysis (LDA) were compared to classify wrist motions using surface electromyogram (EMG). Effect of feature selection in EMG signal processing was also verified by comparing classification accuracy of each feature, and the enhancement of classification accuracy by normalization was confirmed. EMG signals were acquired from two electrodes placed on the forearm of twenty eight healthy subjects and used for recognition of wrist motion. Features were extracted from the obtained EMG signals in the time domain and were applied to classification methods. The difference absolute mean value (DAMV), difference absolute standard deviation value (DASDV), mean absolute value (MAV), root mean square (RMS) were used for composing 16 double features which were combined of two channels. In the classification methods, the highest accuracy of classification showed in the GMM. The most effective combination of classification method and double feature was (MAV, DAMV) of GMM and its classification accuracy was 96.85%. The results of normalization were better than those of non-normalization in GMM, k-NN, and LDA.
基金Supported by Natural Science Foundation of Zhejiang Province of P.R.China(Y104284)
文摘This article explores the ability of multivariate autoregressive model(MAR)and scalar AR model to extract the features from two-lead electrocardiogram signals in order to classify certain cardiac arrhythmias.The classification performance of four different ECG feature sets based on the model coefficients are shown.The data in the analysis including normal sinus rhythm, atria premature contraction,premature ventricular contraction,ventricular tachycardia,ventricular fibrillation and superventricular tachyeardia is obtained from the MIT-BIH database.The classification is performed using a quadratic diacriminant function.The results show the MAR coefficients produce the best results among the four ECG representations and the MAR modeling is a useful classification and diagnosis tool.
基金This project was supported by the High Technology Research and Development Programof China (2002AA103087) .
文摘A model classification rate control method for video coding is proposed. The macro-blocks are classified according to their prediction errors, and different parameters are used in the rate-quantization and distortion-quantization model. The different model parameters are calculated from the previous frame of the same type in the process of coding. These models are used to estimate the relations among rate, distortion and quantization of the current frame. Further steps, such as R-D optimization based quantization adjustment and smoothing of quantization of adjacent macrobloeks, are used to improve the quality. The results of the experiments prove that the technique is effective and can be realized easily. The method presented in the paper can be a good way for MPEG and H. 264 rate control.
文摘The process of human natural scene categorization consists of two correlated stages: visual perception and visual cognition of natural scenes.Inspired by this fact,we propose a biologically plausible approach for natural scene image classification.This approach consists of one visual perception model and two visual cognition models.The visual perception model,composed of two steps,is used to extract discriminative features from natural scene images.In the first step,we mimic the oriented and bandpass properties of human primary visual cortex by a special complex wavelets transform,which can decompose a natural scene image into a series of 2D spatial structure signals.In the second step,a hybrid statistical feature extraction method is used to generate gist features from those 2D spatial structure signals.Then we design a cognitive feedback model to realize adaptive optimization for the visual perception model.At last,we build a multiple semantics based cognition model to imitate human cognitive mode in rapid natural scene categorization.Experiments on natural scene datasets show that the proposed method achieves high efficiency and accuracy for natural scene classification.
文摘[目的/意义]针对实体嵌套、实体类型混淆等问题导致的农业病害命名实体识别(Named Entities Recognition,NER)准确率不高的情况,以PointerNet为基准模型,提出一种基于RoFormer预训练模型的指针网络农业病害NER方法RoFormer-PointerNet。[方法]采用RoFormer预训练模型对输入的文本进行向量化,利用其独特的旋转位置嵌入方法来捕捉位置信息,丰富字词特征信息,从而解决一词多义导致的类型易混淆的问题。使用指针网络进行解码,利用指针网络的首尾指针标注方式抽取句子中的所有实体,首尾指针标注方式可以解决实体抽取中存在的嵌套问题。[结果和讨论]自建农业病害数据集,数据集中包含2867条标注语料,共10282个实体。为验证RoFormer预训练模型在实体抽取上的优越性,采用Word2Vec、BERT、RoBERTa等多种向量化模型进行对比试验,RoFormer-PointerNet与其他模型相比,模型精确率、召回率、F1值均为最优,分别为87.49%,85.76%和86.62%。为验证RoFormer-PointerNet在缓解实体嵌套的优势,与使用最为广泛的双向长短期记忆神经网络(Bidirectional Long Short-Term Memory,BiLSTM)和条件随机场(Conditional Random Field,CRF)模型进行对比试验,RoFormer-PointerNet比RoFormer-BiLSTM模型、RoFormer-CRF模型和RoFormer-BiLSTM-CRF模型分别高出4.8%、5.67%和3.87%,证明用指针网络模型可以很好解决实体嵌套问题。最后验证RoFormer-PointerNet方法在农业病害数据集中的识别性能,针对病害症状、病害名称、防治方法等8类实体进行了识别实验,本方法识别的精确率、召回率和F1值分别为87.49%、85.76%和86.62%,为同类最优。[结论]本研究提出的方法能有效识别中文农业病害文本中的实体,识别效果优于其他模型。在解决实体抽取过程中的实体嵌套和类型混淆等问题方面具有一定优势。
基金supported by National Natural Science Foundation of China(81673480) Project of Urumqi science and Technology Bureau of the Xinjiang Uygur Autonomous Region(Y151310010)
文摘Influenza caused by influenza virus,seriously threaten human life and health.Drug treatment is one of the effective measurement.However,there are only two classes of drugs,one class is M2 blockers and another is neuraminidase(NA) inhibitors.The recent antiviral surveillance studies reported a global significant increase in M2 blocker resistance among influenza viruses,and the resistant virus strains against NA inhibitor are also reported in clinical treatment.Therefore thediscovery of new medicines with low resistance has become very urgent.As all known,traditional medicines with multi-target features and network mechanism often possess low resistance.Compound Yizhihao,which consists of radix isatidis,folium isatidis,Artemisia rupestris,is one of the famous traditional medicine for influenza treatment in China,however its mechanism of action against influenza is unclear.In this study,the multiple targets related with influenza disease and the known chemical constituents from Compound Yizhihao were collected,and multi-target QSAR(mt-QSAR) classification models were developed by Na?e Bayesian algorithm and verified by various datasets.Then the classification models were applied to predict the effective constituents and their drug targets.Finally,the constituent-target-pathway network was constructed,which revealed the effective constituents and their network mechanism in Compound Yizhihao.This study will lay important basis for the clinical uses for influenza treatment and for the further research and development of the effective constituents.
文摘为解决医学文本使用预训练模型处理速度慢、对硬件的算力要求高、难以分辨少样本类别,以及使用传统小模型由于自身限制难以做到足够的准确度的问题,提出了一种融合预训练和元蒸馏的文本分类模型PTMD(fusion of pre-training and meta distillation model)。PTMD针对医学文本的多标签问题,通过对比训练对RoBERTa预训练方法进行微调,再由双向内置注意力简单循环单元充分获取语义信息。最后在传统蒸馏模型的基础上融合元学习和助教的思想,通过教学实验和双级模型等方法,提高模型的教学水平,最终在减少训练成本的基础上得到一个高性能医学文本分类模型。实验结果表明,教师模型在CHIP2019评测三数据集上的F1值达到了85.47%,同时学生模型在F1值损失1.45个百分点的情况下,将模型规模缩小到教师模型的近1/6,效果高于大多数传统预训练模型和知识蒸馏模型,证明了该模型具有良好的实用价值。
