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
One-class support vector machine （OCSVM） and support vector data description （SVDD） are two main domain-based one-class （kernel） classifiers. To reveal their relationship with density estimation in the case of t...One-class support vector machine （OCSVM） and support vector data description （SVDD） are two main domain-based one-class （kernel） classifiers. To reveal their relationship with density estimation in the case of the Gaussian kernel, OCSVM and SVDD are firstly unified into the framework of kernel density estimation, and the essential relationship between them is explicitly revealed. Then the result proves that the density estimation induced by OCSVM or SVDD is in agreement with the true density. Meanwhile, it can also reduce the integrated squared error （ISE）. Finally, experiments on several simulated datasets verify the revealed relationships.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
Turbopump condition monitoring is a significant approach to ensure the safety of liquid rocket engine （LRE）.Because of lack of fault samples,a monitoring system cannot be trained on all possible condition patterns.T...Turbopump condition monitoring is a significant approach to ensure the safety of liquid rocket engine （LRE）.Because of lack of fault samples,a monitoring system cannot be trained on all possible condition patterns.Thus it is important to differentiate abnormal or unknown patterns from normal pattern with novelty detection methods.One-class support vector machine （OCSVM） that has been commonly used for novelty detection cannot deal well with large scale samples.In order to model the normal pattern of the turbopump with OCSVM and so as to monitor the condition of the turbopump,a monitoring method that integrates OCSVM with incremental clustering is presented.In this method,the incremental clustering is used for sample reduction by extracting representative vectors from a large training set.The representative vectors are supposed to distribute uniformly in the object region and fulfill the region.And training OCSVM on these representative vectors yields a novelty detector.By applying this method to the analysis of the turbopump＇s historical test data,it shows that the incremental clustering algorithm can extract 91 representative points from more than 36 000 training vectors,and the OCSVM detector trained on these 91 representative points can recognize spikes in vibration signals caused by different abnormal events such as vane shedding,rub-impact and sensor faults.This monitoring method does not need fault samples during training as classical recognition methods.The method resolves the learning problem of large samples and is an alternative method for condition monitoring of the LRE turbopump.展开更多
microRNAs (miRNAs) are short nucleotide sequences expressed by a genome that are involved in post transcriptional modulation of gene expression. Since miRNAs need to be co-expressed with their target mRNA to observe a...microRNAs (miRNAs) are short nucleotide sequences expressed by a genome that are involved in post transcriptional modulation of gene expression. Since miRNAs need to be co-expressed with their target mRNA to observe an effect and since miRNAs and target interactions can be cooperative, it is currently not possible to develop a comprehensive experimental atlas of miRNAs and their targets. To overcome this limitation, machine learning has been applied to miRNA detection. In general binary learning (two-class) approaches are applied to miRNA discovery. These learners consider both positive (miRNA) and negative (non-miRNA) examples during the training process. One-class classifiers, on the other hand, use only the information for the target class (miRNA). The one-class approach in machine learning is gradually receiving more attention particularly for solving problems where the negative class is not well defined. This is especially true for miRNAs where the positive class can be experimentally confirmed relatively easy, but where it is not currently possible to call any part of a genome a non-miRNA. To do that, it should be co-expressed with all other possible transcripts of the genome, which currently is a futile endeavor. For machine learning, miRNAs need to be transformed into a feature vector and some currently used features like minimum free energy vary widely in the case of plant miRNAs. In this study it was our aim to analyze different methods applying one-class approaches and the effectiveness of motif-based features for prediction of plant miRNA genes. We show that the application of these one-class classifiers is promising and useful for this kind of problem which relies only on sequence- based features such as k-mers and motifs comparing to the results from two-class classification. In some cases the results of one-class are, to our surprise, more accurate than results from two-class classifiers.展开更多
Fuel injectors are considered as an important component of combustion engines. Operational weakness can possibly lead to the complete machine malfunction, decreasing reliability and leading to loss of production. To o...Fuel injectors are considered as an important component of combustion engines. Operational weakness can possibly lead to the complete machine malfunction, decreasing reliability and leading to loss of production. To overcome these circumstances, various condition monitoring techniques can be applied. The application of acoustic signals is common in the field of fault diagnosis of rotating machinery. Advanced signal processing is utilized for the construction of features that are specialized in detecting fuel injector faults. A performance comparison between novelty detection algorithms in the form of one-class classifiers is presented. The one-class classifiers that were tested included One-Class Support Vector Machine (OCSVM) and One-Class Self Organizing Map (OCSOM). The acoustic signals of fuel injectors in different operational conditions were processed for feature extraction. Features from all the signals were used as input to the one-class classifiers. The one-class classifiers were trained only with healthy fuel injector conditions and compared with new experimental data which belonged to different operational conditions that were not included in the training set so as to contribute to generalization. The results present the effectiveness of one-class classifiers for detecting faults in fuel injectors.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
Signature verification is regarded as the most beneficial behavioral characteristic-based biometric feature in security and fraud protection.It is also a popular biometric authentication technology in forensic and com...Signature verification is regarded as the most beneficial behavioral characteristic-based biometric feature in security and fraud protection.It is also a popular biometric authentication technology in forensic and commercial transactions due to its various advantages,including noninvasiveness,user-friendliness,and social and legal acceptability.According to the literature,extensive research has been conducted on signature verification systems in a variety of languages,including English,Hindi,Bangla,and Chinese.However,the Arabic Offline Signature Verification(OSV)system is still a challenging issue that has not been investigated as much by researchers due to the Arabic script being distinguished by changing letter shapes,diacritics,ligatures,and overlapping,making verification more difficult.Recently,signature verification systems have shown promising results for recognizing signatures that are genuine or forgeries;however,performance on skilled forgery detection is still unsatisfactory.Most existing methods require many learning samples to improve verification accuracy,which is a major drawback because the number of available signature samples is often limited in the practical application of signature verification systems.This study addresses these issues by presenting an OSV system based on multifeature fusion and discriminant feature selection using a genetic algorithm(GA).In contrast to existing methods,which use multiclass learning approaches,this study uses a oneclass learning strategy to address imbalanced signature data in the practical application of a signature verification system.The proposed approach is tested on three signature databases(SID)-Arabic handwriting signatures,CEDAR(Center of Excellence for Document Analysis and Recognition),and UTSIG(University of Tehran Persian Signature),and experimental results show that the proposed system outperforms existing systems in terms of reducing the False Acceptance Rate(FAR),False Rejection Rate(FRR),and Equal Error Rate(ERR).The proposed system achieved 5%improvement.展开更多
The application of one-class machine learning is gaining attention in the computational biology community. Different studies have described the use of two-class machine learning to predict microRNAs (miRNAs) gene targ...The application of one-class machine learning is gaining attention in the computational biology community. Different studies have described the use of two-class machine learning to predict microRNAs (miRNAs) gene target. Most of these methods require the generation of an artificial negative class. However, designation of the negative class can be problematic and if it is not properly done can affect the performance of the classifier dramatically and/or yield a biased estimate of performance. We present a study using one-class machine learning for miRNA target discovery and compare one-class to two-class approaches. Of all the one-class methods tested, we found that most of them gave similar accuracy that range from 0.81 to 0.89 while the two-class naive Bayes gave 0.99 accuracy. One and two class methods can both give useful classification accuracies. The advantage of one class methods is that they don’t require any additional effort for choosing the best way of generating the negative class. In these cases one- class methods can be superior to two-class methods when the features which are chosen as representative of that positive class are well defined.展开更多
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.
基金Supported by the National Natural Science Foundation of China(60603029)the Natural Science Foundation of Jiangsu Province(BK2007074)the Natural Science Foundation for Colleges and Universities in Jiangsu Province(06KJB520132)~~
文摘One-class support vector machine （OCSVM） and support vector data description （SVDD） are two main domain-based one-class （kernel） classifiers. To reveal their relationship with density estimation in the case of the Gaussian kernel, OCSVM and SVDD are firstly unified into the framework of kernel density estimation, and the essential relationship between them is explicitly revealed. Then the result proves that the density estimation induced by OCSVM or SVDD is in agreement with the true density. Meanwhile, it can also reduce the integrated squared error （ISE）. Finally, experiments on several simulated datasets verify the revealed relationships.
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 National Natural Science Foundation of China (Grant No. 50675219)Hu’nan Provincial Science Committee Excellent Youth Foundation of China (Grant No. 08JJ1008)
文摘Turbopump condition monitoring is a significant approach to ensure the safety of liquid rocket engine （LRE）.Because of lack of fault samples,a monitoring system cannot be trained on all possible condition patterns.Thus it is important to differentiate abnormal or unknown patterns from normal pattern with novelty detection methods.One-class support vector machine （OCSVM） that has been commonly used for novelty detection cannot deal well with large scale samples.In order to model the normal pattern of the turbopump with OCSVM and so as to monitor the condition of the turbopump,a monitoring method that integrates OCSVM with incremental clustering is presented.In this method,the incremental clustering is used for sample reduction by extracting representative vectors from a large training set.The representative vectors are supposed to distribute uniformly in the object region and fulfill the region.And training OCSVM on these representative vectors yields a novelty detector.By applying this method to the analysis of the turbopump＇s historical test data,it shows that the incremental clustering algorithm can extract 91 representative points from more than 36 000 training vectors,and the OCSVM detector trained on these 91 representative points can recognize spikes in vibration signals caused by different abnormal events such as vane shedding,rub-impact and sensor faults.This monitoring method does not need fault samples during training as classical recognition methods.The method resolves the learning problem of large samples and is an alternative method for condition monitoring of the LRE turbopump.
文摘microRNAs (miRNAs) are short nucleotide sequences expressed by a genome that are involved in post transcriptional modulation of gene expression. Since miRNAs need to be co-expressed with their target mRNA to observe an effect and since miRNAs and target interactions can be cooperative, it is currently not possible to develop a comprehensive experimental atlas of miRNAs and their targets. To overcome this limitation, machine learning has been applied to miRNA detection. In general binary learning (two-class) approaches are applied to miRNA discovery. These learners consider both positive (miRNA) and negative (non-miRNA) examples during the training process. One-class classifiers, on the other hand, use only the information for the target class (miRNA). The one-class approach in machine learning is gradually receiving more attention particularly for solving problems where the negative class is not well defined. This is especially true for miRNAs where the positive class can be experimentally confirmed relatively easy, but where it is not currently possible to call any part of a genome a non-miRNA. To do that, it should be co-expressed with all other possible transcripts of the genome, which currently is a futile endeavor. For machine learning, miRNAs need to be transformed into a feature vector and some currently used features like minimum free energy vary widely in the case of plant miRNAs. In this study it was our aim to analyze different methods applying one-class approaches and the effectiveness of motif-based features for prediction of plant miRNA genes. We show that the application of these one-class classifiers is promising and useful for this kind of problem which relies only on sequence- based features such as k-mers and motifs comparing to the results from two-class classification. In some cases the results of one-class are, to our surprise, more accurate than results from two-class classifiers.
文摘Fuel injectors are considered as an important component of combustion engines. Operational weakness can possibly lead to the complete machine malfunction, decreasing reliability and leading to loss of production. To overcome these circumstances, various condition monitoring techniques can be applied. The application of acoustic signals is common in the field of fault diagnosis of rotating machinery. Advanced signal processing is utilized for the construction of features that are specialized in detecting fuel injector faults. A performance comparison between novelty detection algorithms in the form of one-class classifiers is presented. The one-class classifiers that were tested included One-Class Support Vector Machine (OCSVM) and One-Class Self Organizing Map (OCSOM). The acoustic signals of fuel injectors in different operational conditions were processed for feature extraction. Features from all the signals were used as input to the one-class classifiers. The one-class classifiers were trained only with healthy fuel injector conditions and compared with new experimental data which belonged to different operational conditions that were not included in the training set so as to contribute to generalization. The results present the effectiveness of one-class classifiers for detecting faults in fuel injectors.
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Signature verification is regarded as the most beneficial behavioral characteristic-based biometric feature in security and fraud protection.It is also a popular biometric authentication technology in forensic and commercial transactions due to its various advantages,including noninvasiveness,user-friendliness,and social and legal acceptability.According to the literature,extensive research has been conducted on signature verification systems in a variety of languages,including English,Hindi,Bangla,and Chinese.However,the Arabic Offline Signature Verification(OSV)system is still a challenging issue that has not been investigated as much by researchers due to the Arabic script being distinguished by changing letter shapes,diacritics,ligatures,and overlapping,making verification more difficult.Recently,signature verification systems have shown promising results for recognizing signatures that are genuine or forgeries;however,performance on skilled forgery detection is still unsatisfactory.Most existing methods require many learning samples to improve verification accuracy,which is a major drawback because the number of available signature samples is often limited in the practical application of signature verification systems.This study addresses these issues by presenting an OSV system based on multifeature fusion and discriminant feature selection using a genetic algorithm(GA).In contrast to existing methods,which use multiclass learning approaches,this study uses a oneclass learning strategy to address imbalanced signature data in the practical application of a signature verification system.The proposed approach is tested on three signature databases(SID)-Arabic handwriting signatures,CEDAR(Center of Excellence for Document Analysis and Recognition),and UTSIG(University of Tehran Persian Signature),and experimental results show that the proposed system outperforms existing systems in terms of reducing the False Acceptance Rate(FAR),False Rejection Rate(FRR),and Equal Error Rate(ERR).The proposed system achieved 5%improvement.
文摘The application of one-class machine learning is gaining attention in the computational biology community. Different studies have described the use of two-class machine learning to predict microRNAs (miRNAs) gene target. Most of these methods require the generation of an artificial negative class. However, designation of the negative class can be problematic and if it is not properly done can affect the performance of the classifier dramatically and/or yield a biased estimate of performance. We present a study using one-class machine learning for miRNA target discovery and compare one-class to two-class approaches. Of all the one-class methods tested, we found that most of them gave similar accuracy that range from 0.81 to 0.89 while the two-class naive Bayes gave 0.99 accuracy. One and two class methods can both give useful classification accuracies. The advantage of one class methods is that they don’t require any additional effort for choosing the best way of generating the negative class. In these cases one- class methods can be superior to two-class methods when the features which are chosen as representative of that positive class are well defined.
