为了解决财经微博文本中网民情感状态转移的时序数据分析问题,本文提出一个基于认知情感评价模型(Ortony,Clore&Collins,OCC)和长短期记忆模型(long short term memory,LSTM)的财经微博文本情感分类模型(OCC-LSTM)。基于OCC模型从...为了解决财经微博文本中网民情感状态转移的时序数据分析问题,本文提出一个基于认知情感评价模型(Ortony,Clore&Collins,OCC)和长短期记忆模型(long short term memory,LSTM)的财经微博文本情感分类模型(OCC-LSTM)。基于OCC模型从网民认知角度建立情感规则,对财经微博文本进行情感标注,并作为LSTM模型进行深度学习的训练集;基于LSTM模型,使用深度学习中的TensorFlow框架和Keras模块建立相应的实验模型,进行海量微博数据情感分类,并结合13家上市公司3年的微博文本数据进行实证研究和模型验证对比。实证研究结果发现本文提出的模型取得了89.45%的准确率,高于采用传统的机器学习方式的支持向量机方法 (support vector machine,SVM)和基于深度学习的半监督RAE方法 (semi-supervised recursive auto encoder)。展开更多
Despite the big success of transfer learning techniques in anomaly detection,it is still challenging to achieve good transition of detection rules merely based on the preferred data in the anomaly detection with one-c...Despite the big success of transfer learning techniques in anomaly detection,it is still challenging to achieve good transition of detection rules merely based on the preferred data in the anomaly detection with one-class classification,especially for the data with a large distribution difference.To address this challenge,a novel deep one-class transfer learning algorithm with domain-adversarial training is proposed in this paper.First,by integrating a hypersphere adaptation constraint into domainadversarial neural network,a new hypersphere adversarial training mechanism is designed.Second,an alternative optimization method is derived to seek the optimal network parameters while pushing the hyperspheres built in the source domain and target domain to be as identical as possible.Through transferring oneclass detection rule in the adaptive extraction of domain-invariant feature representation,the end-to-end anomaly detection with one-class classification is then enhanced.Furthermore,a theoretical analysis about the model reliability,as well as the strategy of avoiding invalid and negative transfer,is provided.Experiments are conducted on two typical anomaly detection problems,i.e.,image recognition detection and online early fault detection of rolling bearings.The results demonstrate that the proposed algorithm outperforms the state-of-the-art methods in terms of detection accuracy and robustness.展开更多
MicroRNAs are a class of small, single-stranded RNAs which are produced by non-protein-coding RNA genes with a length of 21-29 nt. They regulate the expression of protein-encoding genes at the post-transcriptional lev...MicroRNAs are a class of small, single-stranded RNAs which are produced by non-protein-coding RNA genes with a length of 21-29 nt. They regulate the expression of protein-encoding genes at the post-transcriptional level and the degradation ofmRNAs by base pairing to mRNAs. Mature miRNAs are processed from 60-90 nt RNA hairpin structures called pre-miRNAs. At present, most of the machine learning computational methods for pre-miRNAs prediction are based on two-class SVM and use structural information of pre-miRNA hairpins. Those methods share a common feature that all of them need a negative dataset in the training dataset and feature selection in both training and testing dataset. In order to avoid selecting false negative examples of miRNA hairpins in the training dataset which may mislead the classifiers, we presented a microRNA prediction algorithm called MirBio based on miRNAs Biogenesis which is trained only on the information of the positive miRNAs class to predict miRNAs. It can predict both pre-miRNAs and miRNAs and get a relatively satisfying result in this study.展开更多
One-class classification problem has become a popular problem in many fields, with a wide range of applications in anomaly detection, fault diagnosis, and face recognition. We investigate the one-class classification ...One-class classification problem has become a popular problem in many fields, with a wide range of applications in anomaly detection, fault diagnosis, and face recognition. We investigate the one-class classification problem for second-order tensor data. Traditional vector-based one-class classification methods such as one-class support vector machine (OCSVM) and least squares one-class support vector machine (LSOCSVM) have limitations when tensor is used as input data, so we propose a new tensor one-class classification method, LSOCSTM, which directly uses tensor as input data. On one hand, using tensor as input data not only enables to classify tensor data, but also for vector data, classifying it after high dimensionalizing it into tensor still improves the classification accuracy and overcomes the over-fitting problem. On the other hand, different from one-class support tensor machine (OCSTM), we use squared loss instead of the original loss function so that we solve a series of linear equations instead of quadratic programming problems. Therefore, we use the distance to the hyperplane as a metric for classification, and the proposed method is more accurate and faster compared to existing methods. The experimental results show the high efficiency of the proposed method compared with several state-of-the-art methods.展开更多
Given the scarcity of Satellite Frequency and Orbit(SFO)resources,it holds paramount importance to establish a comprehensive knowledge graph of SFO field(SFO-KG)and employ knowledge reasoning technology to automatical...Given the scarcity of Satellite Frequency and Orbit(SFO)resources,it holds paramount importance to establish a comprehensive knowledge graph of SFO field(SFO-KG)and employ knowledge reasoning technology to automatically mine available SFO resources.An essential aspect of constructing SFO-KG is the extraction of Chinese entity relations.Unfortunately,there is currently no publicly available Chinese SFO entity Relation Extraction(RE)dataset.Moreover,publicly available SFO text data contain numerous NA(representing for“No Answer”)relation category sentences that resemble other relation sentences and pose challenges in accurate classification,resulting in low recall and precision for the NA relation category in entity RE.Consequently,this issue adversely affects both the accuracy of constructing the knowledge graph and the efficiency of RE processes.To address these challenges,this paper proposes a method for extracting Chinese SFO text entity relations based on dynamic integrated learning.This method includes the construction of a manually annotated Chinese SFO entity RE dataset and a classifier combining features of SFO resource data.The proposed approach combines integrated learning and pre-training models,specifically utilizing Bidirectional Encoder Representation from Transformers(BERT).In addition,it incorporates one-class classification,attention mechanisms,and dynamic feedback mechanisms to improve the performance of the RE model.Experimental results show that the proposed method outperforms the traditional methods in terms of F1 value when extracting entity relations from both balanced and long-tailed datasets.展开更多
Face Presentation Attack Detection(fPAD)plays a vital role in securing face recognition systems against various presentation attacks.While supervised learning-based methods demonstrate effectiveness,they are prone to ...Face Presentation Attack Detection(fPAD)plays a vital role in securing face recognition systems against various presentation attacks.While supervised learning-based methods demonstrate effectiveness,they are prone to overfitting to known attack types and struggle to generalize to novel attack scenarios.Recent studies have explored formulating fPAD as an anomaly detection problem or one-class classification task,enabling the training of generalized models for unknown attack detection.However,conventional anomaly detection approaches encounter difficulties in precisely delineating the boundary between bonafide samples and unknown attacks.To address this challenge,we propose a novel framework focusing on unknown attack detection using exclusively bonafide facial data during training.The core innovation lies in our pseudo-negative sample synthesis(PNSS)strategy,which facilitates learning of compact decision boundaries between bonafide faces and potential attack variations.Specifically,PNSS generates synthetic negative samples within low-likelihood regions of the bonafide feature space to represent diverse unknown attack patterns.To overcome the inherent imbalance between positive and synthetic negative samples during iterative training,we implement a dual-loss mechanism combining focal loss for classification optimization with pairwise confusion loss as a regularizer.This architecture effectively mitigates model bias towards bonafide samples while maintaining discriminative power.Comprehensive evaluations across three benchmark datasets validate the framework’s superior performance.Notably,our PNSS achieves 8%–18% average classification error rate(ACER)reduction compared with state-of-the-art one-class fPAD methods in cross-dataset evaluations on Idiap Replay-Attack and MSU-MFSD datasets.展开更多
In wireless communication,the problem of authenticating the transmitter’s identity is challeng-ing,especially for those terminal devices in which the security schemes based on cryptography are approxi-mately unfeasib...In wireless communication,the problem of authenticating the transmitter’s identity is challeng-ing,especially for those terminal devices in which the security schemes based on cryptography are approxi-mately unfeasible owing to limited resources.In this paper,a physical layer authentication scheme is pro-posed to detect whether there is anomalous access by the attackers disguised as legitimate users.Explicitly,channel state information(CSI)is used as a form of fingerprint to exploit spatial discrimination among de-vices in the wireless network and machine learning(ML)technology is employed to promote the improve-ment of authentication accuracy.Considering that the falsified messages are not accessible for authenticator during the training phase,deep support vector data de-scription(Deep SVDD)is selected to solve the one-class classification(OCC)problem.Simulation results show that Deep SVDD based scheme can tackle the challenges of physical layer authentication in wireless communication environments.展开更多
Laser-induced breakdown spectroscopy(LIBS)can be used for the rapid detection of heavy metal contamination of Tegillarca granosa(T.granosa),but an appropriate classification model needs to be constructed.In the one-cl...Laser-induced breakdown spectroscopy(LIBS)can be used for the rapid detection of heavy metal contamination of Tegillarca granosa(T.granosa),but an appropriate classification model needs to be constructed.In the one-class classification method,only target samples are needed in training process to achieve the recognition of abnormal samples,which is suitable for rapid identification of healthy T.granosa from those contaminated with uncertain heavy metals.The construction of a one-class classification model for heavy metal detection in T.granosa by LIBS has faced the problem of high-dimension and small samples.To solve this problem,a novel one-class classification method was proposed in this study.Here,the principal component scores and the intensity of the residual spectrum were combined as extracted features.Then,a one-class classifier based on Mahalanobis distance using the extracted features was constructed and its threshold was set by leave-one-out crossvalidation.The sensitivity,specificity and accuracy of the proposed method were reached to 1,0.9333 and 0.9667 respectively,which are superior to the previously reported methods.展开更多
To ensure authenticity of sesame oil,an authentication technology was proposed using ion mobility spectrometry(IMS)and chemometrics.One-class classification(OCC)methods including one-class partial least squares(OCPLS)...To ensure authenticity of sesame oil,an authentication technology was proposed using ion mobility spectrometry(IMS)and chemometrics.One-class classification(OCC)methods including one-class partial least squares(OCPLS)and one-class support vector machine(OCSVM)were employed to build authentication models for sesame oil.Subsequently,an independent test set was used to validate the constructed models.Validation set of 45 adulterated oils indicated that prediction correction rate of OCPLS model reached 95.6%(43 out of 45).Moreover,the complete set of sesame oils adulterated by sesame oil essence could be identified as counterfeit.Compared with previous studies,OCPLS model could work to identify untargeted adulteration.In conclusion,OCC method could effectively detect adulterated sesame oils containing as little as 10%other vegetable oils.This study provided a rapid screening method for adulterated sesame oil in market surveillance and a reference for developing authentication methods of other edible oils.展开更多
One-class support vector machines （one-class SVMs） are powerful tools that are widely used in many applications. This paper describes a semi-supervised one-class SVM that uses supervision in terms of relative compar...One-class support vector machines （one-class SVMs） are powerful tools that are widely used in many applications. This paper describes a semi-supervised one-class SVM that uses supervision in terms of relative comparisons. The analysis uses a hypersphere version of one-class SVMs with a penalty term appended to the objective function. The method simultaneously finds the minimum sphere in the feature space that encloses most of the target points and considers the relative comparisons. The result is a standard convex quadratic programming problem, which can be solved by adapting standard methods for SVM training, i.e., sequential minimal optimization. This one-class SVM can be applied to semi-supervised clustering and multi-classification problems. Tests show that this method achieves higher accuracy and better generalization performance than previous SVMs.展开更多
Keystroke dynamics is the process to identify or authenticate individuals based on their typing rhythm behaviors. Several classifications have been proposed to verify a user＇s legitimacy, and the performances of thes...Keystroke dynamics is the process to identify or authenticate individuals based on their typing rhythm behaviors. Several classifications have been proposed to verify a user＇s legitimacy, and the performances of these classifications should be confirmed to identify the most promising research direction. However, classification research contains several experiments with different conditions such as datasets and methodologies. This study aims to benchmark the algorithms to the same dataset and features to equally measure all performances. Using a dataset that contains the typing rhythm of 51 subjects, we implement and evaluate 15 classifiers measured by Fl-measure, which is the harmonic mean of a false-negative identification rate and false-positive identification rate. We also develop a methodology to process the typing data. By considering a case in which the model will reject the outsider, we tested the algorithms on an open set. Additionally, we tested different parameters in random forest and k nearest neighbors classifications to achieve better results and explore the cause of their high performance. We also tested the dataset on one-class classification and explained the results of the experiment. The top-performing classifier achieves an Fl-measure rate of 92% while using the normalized typing data of 50 subjects to train and the remaining data to test. The results, along with the normalization methodology, constitute a benchmark for comparing the classifiers and measuring the performance of keystroke dynamics for insider detection.展开更多
Objectives:This study presents a method combining a one-class classifier and laser-induced breakdown spectrometry(LIBS)to quickly identify healthy Tegillarca granosa(T.granosa).Materials and Methods:The sum of ranking...Objectives:This study presents a method combining a one-class classifier and laser-induced breakdown spectrometry(LIBS)to quickly identify healthy Tegillarca granosa(T.granosa).Materials and Methods:The sum of ranking differences(SRD)was used to fuse multiple anomaly detection metrics to build the one-class classifier,which was only trained with healthy T.granosa.The one-class classifier can identify healthy T.granosa to exclude non-healthy T.granosa.The proposed method calculated multiple anomaly detection metrics and standardized them to obtain a fusion matrix.Based on the fusion matrix,the samples were ranked by SRD and those ranked lowest and below the threshold were considered to be unhealthy.Results:Multiple anomaly detection metrics were fused by the SRD algorithm and tested on each band,and the final fusion model achieved an accuracy rate of 98.46%,a sensitivity of 100%,and a specificity of 80%.The remaining three single classification models obtained the following results:the SVDD model achieved an accuracy rate of 87.69%,a sensitivity of 90%,and a specificity of 60%;the OCSVM model achieved an accuracy rate of 80%,a sensitivity of 76.67%,and a specificity of 60%;and the DD-SIMCA model achieved an accuracy rate of 95.38%,a sensitivity of 98.33%,and a specificity of 60%.Conclusions:The experimental results showed that the proposed method achieved better results than the traditional one-class classification methods with a single metric.Therefore,the fusion method effectively improves the performance of traditional one-class classifiers when using LIBS to quickly identify healthy substances(healthy T.granosa).展开更多
文摘为了解决财经微博文本中网民情感状态转移的时序数据分析问题,本文提出一个基于认知情感评价模型(Ortony,Clore&Collins,OCC)和长短期记忆模型(long short term memory,LSTM)的财经微博文本情感分类模型(OCC-LSTM)。基于OCC模型从网民认知角度建立情感规则,对财经微博文本进行情感标注,并作为LSTM模型进行深度学习的训练集;基于LSTM模型,使用深度学习中的TensorFlow框架和Keras模块建立相应的实验模型,进行海量微博数据情感分类,并结合13家上市公司3年的微博文本数据进行实证研究和模型验证对比。实证研究结果发现本文提出的模型取得了89.45%的准确率,高于采用传统的机器学习方式的支持向量机方法 (support vector machine,SVM)和基于深度学习的半监督RAE方法 (semi-supervised recursive auto encoder)。
基金supported by the National Natural Science Foundation of China(NSFC)(U1704158)Henan Province Technologies Research and Development Project of China(212102210103)+1 种基金the NSFC Development Funding of Henan Normal University(2020PL09)the University of Manitoba Research Grants Program(URGP)。
文摘Despite the big success of transfer learning techniques in anomaly detection,it is still challenging to achieve good transition of detection rules merely based on the preferred data in the anomaly detection with one-class classification,especially for the data with a large distribution difference.To address this challenge,a novel deep one-class transfer learning algorithm with domain-adversarial training is proposed in this paper.First,by integrating a hypersphere adaptation constraint into domainadversarial neural network,a new hypersphere adversarial training mechanism is designed.Second,an alternative optimization method is derived to seek the optimal network parameters while pushing the hyperspheres built in the source domain and target domain to be as identical as possible.Through transferring oneclass detection rule in the adaptive extraction of domain-invariant feature representation,the end-to-end anomaly detection with one-class classification is then enhanced.Furthermore,a theoretical analysis about the model reliability,as well as the strategy of avoiding invalid and negative transfer,is provided.Experiments are conducted on two typical anomaly detection problems,i.e.,image recognition detection and online early fault detection of rolling bearings.The results demonstrate that the proposed algorithm outperforms the state-of-the-art methods in terms of detection accuracy and robustness.
基金Supported by the National Natural Science Foundation of China(No.60971089)
文摘MicroRNAs are a class of small, single-stranded RNAs which are produced by non-protein-coding RNA genes with a length of 21-29 nt. They regulate the expression of protein-encoding genes at the post-transcriptional level and the degradation ofmRNAs by base pairing to mRNAs. Mature miRNAs are processed from 60-90 nt RNA hairpin structures called pre-miRNAs. At present, most of the machine learning computational methods for pre-miRNAs prediction are based on two-class SVM and use structural information of pre-miRNA hairpins. Those methods share a common feature that all of them need a negative dataset in the training dataset and feature selection in both training and testing dataset. In order to avoid selecting false negative examples of miRNA hairpins in the training dataset which may mislead the classifiers, we presented a microRNA prediction algorithm called MirBio based on miRNAs Biogenesis which is trained only on the information of the positive miRNAs class to predict miRNAs. It can predict both pre-miRNAs and miRNAs and get a relatively satisfying result in this study.
文摘One-class classification problem has become a popular problem in many fields, with a wide range of applications in anomaly detection, fault diagnosis, and face recognition. We investigate the one-class classification problem for second-order tensor data. Traditional vector-based one-class classification methods such as one-class support vector machine (OCSVM) and least squares one-class support vector machine (LSOCSVM) have limitations when tensor is used as input data, so we propose a new tensor one-class classification method, LSOCSTM, which directly uses tensor as input data. On one hand, using tensor as input data not only enables to classify tensor data, but also for vector data, classifying it after high dimensionalizing it into tensor still improves the classification accuracy and overcomes the over-fitting problem. On the other hand, different from one-class support tensor machine (OCSTM), we use squared loss instead of the original loss function so that we solve a series of linear equations instead of quadratic programming problems. Therefore, we use the distance to the hyperplane as a metric for classification, and the proposed method is more accurate and faster compared to existing methods. The experimental results show the high efficiency of the proposed method compared with several state-of-the-art methods.
文摘Given the scarcity of Satellite Frequency and Orbit(SFO)resources,it holds paramount importance to establish a comprehensive knowledge graph of SFO field(SFO-KG)and employ knowledge reasoning technology to automatically mine available SFO resources.An essential aspect of constructing SFO-KG is the extraction of Chinese entity relations.Unfortunately,there is currently no publicly available Chinese SFO entity Relation Extraction(RE)dataset.Moreover,publicly available SFO text data contain numerous NA(representing for“No Answer”)relation category sentences that resemble other relation sentences and pose challenges in accurate classification,resulting in low recall and precision for the NA relation category in entity RE.Consequently,this issue adversely affects both the accuracy of constructing the knowledge graph and the efficiency of RE processes.To address these challenges,this paper proposes a method for extracting Chinese SFO text entity relations based on dynamic integrated learning.This method includes the construction of a manually annotated Chinese SFO entity RE dataset and a classifier combining features of SFO resource data.The proposed approach combines integrated learning and pre-training models,specifically utilizing Bidirectional Encoder Representation from Transformers(BERT).In addition,it incorporates one-class classification,attention mechanisms,and dynamic feedback mechanisms to improve the performance of the RE model.Experimental results show that the proposed method outperforms the traditional methods in terms of F1 value when extracting entity relations from both balanced and long-tailed datasets.
基金supported in part by the National Natural Science Foundation of China under Grants 61972267,and 61772070in part by the Natural Science Foundation of Hebei Province under Grant F2024210005.
文摘Face Presentation Attack Detection(fPAD)plays a vital role in securing face recognition systems against various presentation attacks.While supervised learning-based methods demonstrate effectiveness,they are prone to overfitting to known attack types and struggle to generalize to novel attack scenarios.Recent studies have explored formulating fPAD as an anomaly detection problem or one-class classification task,enabling the training of generalized models for unknown attack detection.However,conventional anomaly detection approaches encounter difficulties in precisely delineating the boundary between bonafide samples and unknown attacks.To address this challenge,we propose a novel framework focusing on unknown attack detection using exclusively bonafide facial data during training.The core innovation lies in our pseudo-negative sample synthesis(PNSS)strategy,which facilitates learning of compact decision boundaries between bonafide faces and potential attack variations.Specifically,PNSS generates synthetic negative samples within low-likelihood regions of the bonafide feature space to represent diverse unknown attack patterns.To overcome the inherent imbalance between positive and synthetic negative samples during iterative training,we implement a dual-loss mechanism combining focal loss for classification optimization with pairwise confusion loss as a regularizer.This architecture effectively mitigates model bias towards bonafide samples while maintaining discriminative power.Comprehensive evaluations across three benchmark datasets validate the framework’s superior performance.Notably,our PNSS achieves 8%–18% average classification error rate(ACER)reduction compared with state-of-the-art one-class fPAD methods in cross-dataset evaluations on Idiap Replay-Attack and MSU-MFSD datasets.
基金partially supported by the National Key Research and Development Project under Grant2020YFB1806805Social Development Projects of Jiangsu Science and Technology Department under Grant No.BE2018704
文摘In wireless communication,the problem of authenticating the transmitter’s identity is challeng-ing,especially for those terminal devices in which the security schemes based on cryptography are approxi-mately unfeasible owing to limited resources.In this paper,a physical layer authentication scheme is pro-posed to detect whether there is anomalous access by the attackers disguised as legitimate users.Explicitly,channel state information(CSI)is used as a form of fingerprint to exploit spatial discrimination among de-vices in the wireless network and machine learning(ML)technology is employed to promote the improve-ment of authentication accuracy.Considering that the falsified messages are not accessible for authenticator during the training phase,deep support vector data de-scription(Deep SVDD)is selected to solve the one-class classification(OCC)problem.Simulation results show that Deep SVDD based scheme can tackle the challenges of physical layer authentication in wireless communication environments.
基金supported by the Zhejiang Natural Science Foundation of China(Grant No.LY21C200001,LY20F030019)National Natural Science Foundation of China(Grant No.62105245,62071386)+1 种基金Wenzhou Major Scientific and Technological Innovation Projects of China(Grant No.ZG2021029,ZY2021027)the Wenzhou Science and Technology Bureau General Project(Grant No.S2020011).
文摘Laser-induced breakdown spectroscopy(LIBS)can be used for the rapid detection of heavy metal contamination of Tegillarca granosa(T.granosa),but an appropriate classification model needs to be constructed.In the one-class classification method,only target samples are needed in training process to achieve the recognition of abnormal samples,which is suitable for rapid identification of healthy T.granosa from those contaminated with uncertain heavy metals.The construction of a one-class classification model for heavy metal detection in T.granosa by LIBS has faced the problem of high-dimension and small samples.To solve this problem,a novel one-class classification method was proposed in this study.Here,the principal component scores and the intensity of the residual spectrum were combined as extracted features.Then,a one-class classifier based on Mahalanobis distance using the extracted features was constructed and its threshold was set by leave-one-out crossvalidation.The sensitivity,specificity and accuracy of the proposed method were reached to 1,0.9333 and 0.9667 respectively,which are superior to the previously reported methods.
基金This work was supported by the National Science and Technology Major Project of China(2017YFC1601700)the National Nature Foundation Committee of P.R.China(31871886)+2 种基金the National Major Project for Agro-product Quality&Safety Risk Assessment(GJFP2019003)the earmarked fund for China Agricultural Research System(CARS-12)the Fundamental Research Funds for Central Non-profit Scientific Institution(1610172018002 and 1610172018012).
文摘To ensure authenticity of sesame oil,an authentication technology was proposed using ion mobility spectrometry(IMS)and chemometrics.One-class classification(OCC)methods including one-class partial least squares(OCPLS)and one-class support vector machine(OCSVM)were employed to build authentication models for sesame oil.Subsequently,an independent test set was used to validate the constructed models.Validation set of 45 adulterated oils indicated that prediction correction rate of OCPLS model reached 95.6%(43 out of 45).Moreover,the complete set of sesame oils adulterated by sesame oil essence could be identified as counterfeit.Compared with previous studies,OCPLS model could work to identify untargeted adulteration.In conclusion,OCC method could effectively detect adulterated sesame oils containing as little as 10%other vegetable oils.This study provided a rapid screening method for adulterated sesame oil in market surveillance and a reference for developing authentication methods of other edible oils.
基金Supported by the National Natural Science Foundation of China(No. 60872070)
文摘One-class support vector machines （one-class SVMs） are powerful tools that are widely used in many applications. This paper describes a semi-supervised one-class SVM that uses supervision in terms of relative comparisons. The analysis uses a hypersphere version of one-class SVMs with a penalty term appended to the objective function. The method simultaneously finds the minimum sphere in the feature space that encloses most of the target points and considers the relative comparisons. The result is a standard convex quadratic programming problem, which can be solved by adapting standard methods for SVM training, i.e., sequential minimal optimization. This one-class SVM can be applied to semi-supervised clustering and multi-classification problems. Tests show that this method achieves higher accuracy and better generalization performance than previous SVMs.
基金supported by the National Natural Science Foundation of China (Nos. 61403301 and 61773310)the China Postdoctoral Science Foundation (Nos. 2014M560783 and 2015T81032)+1 种基金the Natural Science Foundation of Shaanxi Province (No. 2015JQ6216)the Fundamental Research Funds for the Central Universities (No. xjj2015115)
文摘Keystroke dynamics is the process to identify or authenticate individuals based on their typing rhythm behaviors. Several classifications have been proposed to verify a user＇s legitimacy, and the performances of these classifications should be confirmed to identify the most promising research direction. However, classification research contains several experiments with different conditions such as datasets and methodologies. This study aims to benchmark the algorithms to the same dataset and features to equally measure all performances. Using a dataset that contains the typing rhythm of 51 subjects, we implement and evaluate 15 classifiers measured by Fl-measure, which is the harmonic mean of a false-negative identification rate and false-positive identification rate. We also develop a methodology to process the typing data. By considering a case in which the model will reject the outsider, we tested the algorithms on an open set. Additionally, we tested different parameters in random forest and k nearest neighbors classifications to achieve better results and explore the cause of their high performance. We also tested the dataset on one-class classification and explained the results of the experiment. The top-performing classifier achieves an Fl-measure rate of 92% while using the normalized typing data of 50 subjects to train and the remaining data to test. The results, along with the normalization methodology, constitute a benchmark for comparing the classifiers and measuring the performance of keystroke dynamics for insider detection.
基金The authors would like to acknowledge the financial support provided by the Natural Science Foundation of Zhejiang(No.LY21C200001)China,the National Natural Science Foundation of China(Nos.62105245 and 61805180)the Wenzhou Science and Technology Bureau General Project(Nos.S2020011 and G20200044),China。
文摘Objectives:This study presents a method combining a one-class classifier and laser-induced breakdown spectrometry(LIBS)to quickly identify healthy Tegillarca granosa(T.granosa).Materials and Methods:The sum of ranking differences(SRD)was used to fuse multiple anomaly detection metrics to build the one-class classifier,which was only trained with healthy T.granosa.The one-class classifier can identify healthy T.granosa to exclude non-healthy T.granosa.The proposed method calculated multiple anomaly detection metrics and standardized them to obtain a fusion matrix.Based on the fusion matrix,the samples were ranked by SRD and those ranked lowest and below the threshold were considered to be unhealthy.Results:Multiple anomaly detection metrics were fused by the SRD algorithm and tested on each band,and the final fusion model achieved an accuracy rate of 98.46%,a sensitivity of 100%,and a specificity of 80%.The remaining three single classification models obtained the following results:the SVDD model achieved an accuracy rate of 87.69%,a sensitivity of 90%,and a specificity of 60%;the OCSVM model achieved an accuracy rate of 80%,a sensitivity of 76.67%,and a specificity of 60%;and the DD-SIMCA model achieved an accuracy rate of 95.38%,a sensitivity of 98.33%,and a specificity of 60%.Conclusions:The experimental results showed that the proposed method achieved better results than the traditional one-class classification methods with a single metric.Therefore,the fusion method effectively improves the performance of traditional one-class classifiers when using LIBS to quickly identify healthy substances(healthy T.granosa).
