The Experts below are selected from a list of 127716 Experts worldwide ranked by ideXlab platform
Phengann Heng - One of the best experts on this subject based on the ideXlab platform.
-
sparse Support Matrix machine
Pattern Recognition, 2018Co-Authors: Qingqing Zheng, Fengyuan Zhu, Jing Qin, Badong Chen, Phengann HengAbstract:Abstract Modern technologies have been producing data with complex intrinsic structures, which can be naturally represented as two-dimensional matrices, such as gray digital images, and electroencephalography (EEG) signals. When processing these data for classification, traditional classifiers, such as Support vector machine (SVM) and logistic regression, have to reshape each input Matrix into a feature vector, resulting in the loss of structural information. In contrast, modern classification methods such as Support Matrix machine capture these structures by regularizing the regression Matrix to be low-rank. These methods assume that all entities within each input Matrix can serve as the explanatory features for its label. However, in real-world applications, many features are redundant and useless for certain classification tasks, thus it is important to perform feature selection to filter out redundant features for more interpretable modeling. In this paper, we tackle this issue, and propose a novel classification technique called Sparse Support Matrix Machine (SSMM), which is favored for taking both the intrinsic structure of each input Matrix and feature selection into consideration simultaneously. The proposed SSMM is defined as a hinge loss for model fitting, with a new regularization on the regression Matrix. Specifically, the new regularization term is a linear combination of nuclear norm and l1 norm, to consider the low-rank property and sparse property respectively. The resulting optimization problem is convex, and motivates us to propose a novel and efficient generalized forward-backward algorithm for solving it. To evaluate the effectiveness of our method, we conduct comparative studies on the applications of both image and EEG data classification problems. Our approach achieves state-of-the-art performance consistently. It shows the promise of our SSMM method on real-world applications.
-
multiclass Support Matrix machine for single trial eeg classification
Neurocomputing, 2018Co-Authors: Qingqing Zheng, Fengyuan Zhu, Jing Qin, Phengann HengAbstract:Abstract We propose a novel multiclass classifier for single trial electroencephalogram (EEG) data in Matrix form, namely multiclass Support Matrix machine (MSMM), aiming at improving the classification accuracy of multiclass EEG signals, and hence enhancing the performance of EEG-based brain computer interfaces (BCIs) involving multiple mental activities. In order to construct the MSMM, we propose a novel objective function, which is composed of a multiclass hinge loss term and a combined regularization term. We first formulate the multiclass hinge loss by extending the margin rescaling loss to Support Matrix-form data. We then devise the regularization term by combining the squared Frobenius norm of tensor-form model parameter and the nuclear norm of Matrix-form hyperplanes extracted from the model parameter. While the Frobenius norm prevents over-fitting when training the model, the nuclear norm captures the structural information within the Matrix data. We further propose an efficient solver for MSMM based on the alternating direction method of multipliers (ADMM) framework. We conduct extensive experiments on two benchmark EEG datasets. Experimental results show that MSMM achieves much better performance than state-of-the-art classifiers and yields a mean kappa value of 0.880 and 0.648 for dataset IIIa of BCI III and dataset IIa of BCI IV, respectively. To our best knowledge, MSMM is the first classifier that Supports multiclass classification for Matrix-form EEG data. The proposed MSMM enables easier and more efficient implementation of robust multi-task BCIs, and therefore has potential to promote the wider use of BCI technology.
-
robust Support Matrix machine for single trial eeg classification
International Conference of the IEEE Engineering in Medicine and Biology Society, 2018Co-Authors: Qingqing Zheng, Fengyuan Zhu, Phengann HengAbstract:Electroencephalogram (EEG) signals are of complex structure and can be naturally represented as matrices. Classification is one of the most important steps for EEG signal processing. Newly developed classifiers can handle these Matrix-form data by adding low-rank constraint to leverage the correlation within each data. However, classification of EEG signals is still challenging, because EEG signals are always contaminated by measurement artifacts, outliers, and non-standard noise sources. As a result, existing Matrix classifiers may suffer from performance degradation, because they typically assume that the input EEG signals are clean. In this paper, to account for intra-sample outliers, we propose a novel classifier called a robust Support Matrix machine (RSMM), for single trial EEG data in Matrix form. Inspired by the fact that empirical EEG signals contain strong correlation information, we assume that each EEG Matrix can be decomposed into a latent low-rank clean Matrix plus a sparse noise Matrix. We simultaneously perform signal recovery and train the classifier based on the clean EEG matrices. We formulate our RSMM in a unified framework and present an effective solver based on the alternating direction method of multipliers. To evaluate the proposed method, we conduct extensive classification experiments on real binary EEG signals. The experimental results show that our method has outperformed the state-of-the-art Matrix classifiers. This paper may lead to the development of robust brain–computer interfaces (BCIs) with intuitive motor imagery and thus promote the broad use of the noninvasive BCIs technology.
Haiyang Pan - One of the best experts on this subject based on the ideXlab platform.
-
an intelligent fault diagnosis method for roller bearing using symplectic hyperdisk Matrix machine
Applied Soft Computing, 2021Co-Authors: Haiyang Pan, Jinde ZhengAbstract:Abstract Support Matrix machine (SMM) is a new and effective classification method, which has been applied in the field of image processing. In this paper, an improved SMM called symplectic hyperdisk Matrix machine (SHMM) is proposed and applied to the roller bearing fault diagnosis. In SHMM, the symplectic geometry similarity transformation (SGST) is used to obtain the dimensionless feature Matrix, which protects the signal structure information while weakening the interference of noise. Then, different types of hyperdisks are constructed to divide different types of data, several more realistic hyperdisk prediction models can be obtained and the problem of under estimation is avoided. In order to fully mine the spatial structure information, the feature Matrix is mapped to the high-dimensional space by kernel technology, and the decision function is established by using the structure information hidden of the input Matrix in the SHMM. Experimental results of three datasets of roller bearing show that, compared with symplectic geometry Matrix machine (SGMM), SMM, Support vector machine (SVM) and radial basis function (RBF) neural network methods, the proposed SHMM has good application effect in roller bearing fault diagnosis.
-
symplectic incremental Matrix machine and its application in roller bearing condition monitoring
Applied Soft Computing, 2020Co-Authors: Yu Yang, Haiyang Pan, Ping Wang, Jian Wang, Junsheng ChengAbstract:Abstract For roller bearing condition monitoring, the collected signals have complex internal structure, which can be naturally represented as matrices. Support Matrix machine (SMM), as a new classifier with matrices as inputs, makes full use of the correlation between rows and columns of matrices and achieves ideal classification results. Unfortunately, SMM have ignored the issue of redundant features, which seriously affects the operational efficiency and recognition accuracy of algorithm. In this paper, we introduce symplectic geometry, l 1 -norm and incremental proximal descent (IPD) to SMM, and symplectic incremental Matrix machine (SIMM) is proposed. In SIMM, through symplectic geometry similarity transformation, the de-noising symplectic geometry coefficient Matrix is obtained, and the noise robustness of SMM method is therefore improved. Moreover, l 1 -norm is used to constrain the objective function, which can weaken the influence of redundant features, and thus greatly improving the recognition accuracy of SMM. Meanwhile, we use IPD to solve the objective function, which can obviously enhance the algorithm efficiency under the constant recognition rates. The experimental results of two kinds of roller bearings show that the proposed method has a good effectiveness in roller bearing condition monitoring, and the achieved recognition rate can reach 3%–25% much higher than those of the traditional recognition methods in 5-cross validation.
-
symplectic interactive Support Matrix machine and its application in roller bearing condition monitoring
Neurocomputing, 2020Co-Authors: Yu Yang, Haiyang Pan, Jinde Zheng, Junsheng ChengAbstract:Abstract Support Matrix machine (SMM) is an effective method to solve the problem of mechanical condition monitoring while the Matrix is taken as the input. It makes full use of the effective information between rows and columns of the Matrix to establish an ideal prediction model and achieve good condition monitoring results. However, Similar to Support vector machine (SVM), the core principle of the SMM is to distinguish the data effectively by two parallel hyperplanes. Unfortunately, two parallel hyperplanes may not be able to maximize the interval. Therefore, the concept of interactive Support Matrix machine (ISMM) is proposed, which constructs a pair of interactive hyperplanes to maximize the interval between two types of data. Interactive hyperplanes may be more able to distinguish between two types of data, so that each hyperplane is as close as to one of the two types and as far away as possible from the other. However, the input of the model often contain noise information, which seriously interferes with the classification results. Therefore, a symplectic interactive Support Matrix machine (SISMM) method is further proposed, which combines symplectic geometry similarity transformation (SGST) with ISMM. In SISMM, it can directly get the symplectic geometry coefficient Matrix without noise from the original signal, and intelligent classification recognition is realized. By analyzing and comparing the signal of roller bearings, the results show that the proposed method has better recognition performance and it is feasible for roller bearing condition monitoring.
-
a fault diagnosis approach for roller bearing based on symplectic geometry Matrix machine
Mechanism and Machine Theory, 2019Co-Authors: Yu Yang, Haiyang Pan, Jinde Zheng, Junsheng ChengAbstract:Abstract In many classification problems such as roller bearing fault diagnosis, it is often met that input samples are two-dimensional matrices constructed by vibration signals, and the rows or columns in the input matrices are strongly correlated. Support Matrix machine (SMM) is a new classifier with Matrix as input, which has a good diagnostic effect by using of Matrix structural information. Unfortunately, SMM algorithm is essentially binary, which need carry on the multiple binary classifications for multi-class classification problem. Meanwhile, SMM method has limitations in dealing with the complex input matrices, such as noise robustness and convergence problem. Therefore, a new classification method, called symplectic geometry Matrix machine (SGMM), is proposed in this paper. In SGMM, by using symplectic geometry similarity transformation, the proposed method not only protects the original structure of the signal, but also automatically extracts noiseless features to establish weight coefficient model, which can achieve multi-class tasks. Meanwhile, because of establishment of weight coefficient model, the convergence problem can be avoided. The roller bearing fault signals are used to demonstrate the validity of the SGMM method, and the analysis results indicate that the proposed method has a good effectiveness in roller bearing fault diagnosis.
Kenton W Gregory - One of the best experts on this subject based on the ideXlab platform.
-
mechanical property characterization of electrospun recombinant human tropoelastin for vascular graft biomaterials
Acta Biomaterialia, 2012Co-Authors: Kathryn A Mckenna, Monica T Hinds, Rebecca C Sarao, Ping Cheng Wu, Cheryl L Maslen, Robert W Glanville, Darcie Babcock, Kenton W GregoryAbstract:The development of vascular grafts has focused on finding a biomaterial that is non-thrombogenic, minimizes intimal hyperplasia, matches the mechanical properties of native vessels and allows for regeneration of arterial tissue. In this study, the structural and mechanical properties and the vascular cell compatibility of electrospun recombinant human tropoelastin (rTE) were evaluated as a potential vascular graft Support Matrix. Disuccinimidyl suberate (DSS) was used to cross-link electrospun rTE fibers to produce a polymeric recombinant tropoelastin (prTE) Matrix that is stable in aqueous environments. Tubular 1 cm diameter prTE samples were constructed for uniaxial tensile testing and 4 mm small-diameter prTE tubular scaffolds were produced for burst pressure and cell compatibility evaluations from 15 wt% rTE solutions. Uniaxial tensile tests demonstrated an average ultimate tensile strength (UTS) of 0.36±0.05 MPa and elastic moduli of 0.15±0.04 MPa and 0.91±0.16 MPa, which were comparable to extracted native elastin. Burst pressures of 485 ± 25 mmHg were obtained from 4 mm ID scaffolds with 453 ± 74 μm average wall thickness. prTE Supported endothelial cell growth with typical endothelial cell cobblestone morphology after 48 hours in culture. Cross-linked electrospun recombinant human tropoelastin has promising properties for utilization as a vascular graft biomaterial with customizable dimensions, a compliant Matrix, and vascular cell compatibility.
-
mechanical property characterization of electrospun recombinant human tropoelastin for vascular graft biomaterials
Acta Biomaterialia, 2012Co-Authors: Kathryn A Mckenna, Monica T Hinds, Rebecca C Sarao, Cheryl L Maslen, Robert W Glanville, Darcie Babcock, Kenton W GregoryAbstract:The development of vascular grafts has focused on finding a biomaterial that is non-thrombogenic, minimizes intimal hyperplasia, matches the mechanical properties of native vessels and allows for regeneration of arterial tissue. In this study, the structural and mechanical properties and the vascular cell compatibility of electrospun recombinant human tropoelastin (rTE) were evaluated as a potential vascular graft Support Matrix. Disuccinimidyl suberate (DSS) was used to cross-link electrospun rTE fibers to produce a polymeric recombinant tropoelastin (prTE) Matrix that is stable in aqueous environments. Tubular 1cm diameter prTE samples were constructed for uniaxial tensile testing and 4mm small-diameter prTE tubular scaffolds were produced for burst pressure and cell compatibility evaluations from 15 wt.% rTE solutions. Uniaxial tensile tests demonstrated an average ultimate tensile strength (UTS) of 0.36±0.05 MPa and elastic moduli of 0.15±0.04 and 0.91±0.16 MPa, which were comparable to extracted native elastin. Burst pressures of 485±25 mm Hg were obtained from 4mm internal diameter scaffolds with 453±74 μm average wall thickness. prTE Supported endothelial cell growth with typical endothelial cell cobblestone morphology after 48 h in culture. Cross-linked electrospun rTE has promising properties for utilization as a vascular graft biomaterial with customizable dimensions, a compliant Matrix and vascular cell compatibility.
Qingqing Zheng - One of the best experts on this subject based on the ideXlab platform.
-
sparse Support Matrix machine
Pattern Recognition, 2018Co-Authors: Qingqing Zheng, Fengyuan Zhu, Jing Qin, Badong Chen, Phengann HengAbstract:Abstract Modern technologies have been producing data with complex intrinsic structures, which can be naturally represented as two-dimensional matrices, such as gray digital images, and electroencephalography (EEG) signals. When processing these data for classification, traditional classifiers, such as Support vector machine (SVM) and logistic regression, have to reshape each input Matrix into a feature vector, resulting in the loss of structural information. In contrast, modern classification methods such as Support Matrix machine capture these structures by regularizing the regression Matrix to be low-rank. These methods assume that all entities within each input Matrix can serve as the explanatory features for its label. However, in real-world applications, many features are redundant and useless for certain classification tasks, thus it is important to perform feature selection to filter out redundant features for more interpretable modeling. In this paper, we tackle this issue, and propose a novel classification technique called Sparse Support Matrix Machine (SSMM), which is favored for taking both the intrinsic structure of each input Matrix and feature selection into consideration simultaneously. The proposed SSMM is defined as a hinge loss for model fitting, with a new regularization on the regression Matrix. Specifically, the new regularization term is a linear combination of nuclear norm and l1 norm, to consider the low-rank property and sparse property respectively. The resulting optimization problem is convex, and motivates us to propose a novel and efficient generalized forward-backward algorithm for solving it. To evaluate the effectiveness of our method, we conduct comparative studies on the applications of both image and EEG data classification problems. Our approach achieves state-of-the-art performance consistently. It shows the promise of our SSMM method on real-world applications.
-
multiclass Support Matrix machine for single trial eeg classification
Neurocomputing, 2018Co-Authors: Qingqing Zheng, Fengyuan Zhu, Jing Qin, Phengann HengAbstract:Abstract We propose a novel multiclass classifier for single trial electroencephalogram (EEG) data in Matrix form, namely multiclass Support Matrix machine (MSMM), aiming at improving the classification accuracy of multiclass EEG signals, and hence enhancing the performance of EEG-based brain computer interfaces (BCIs) involving multiple mental activities. In order to construct the MSMM, we propose a novel objective function, which is composed of a multiclass hinge loss term and a combined regularization term. We first formulate the multiclass hinge loss by extending the margin rescaling loss to Support Matrix-form data. We then devise the regularization term by combining the squared Frobenius norm of tensor-form model parameter and the nuclear norm of Matrix-form hyperplanes extracted from the model parameter. While the Frobenius norm prevents over-fitting when training the model, the nuclear norm captures the structural information within the Matrix data. We further propose an efficient solver for MSMM based on the alternating direction method of multipliers (ADMM) framework. We conduct extensive experiments on two benchmark EEG datasets. Experimental results show that MSMM achieves much better performance than state-of-the-art classifiers and yields a mean kappa value of 0.880 and 0.648 for dataset IIIa of BCI III and dataset IIa of BCI IV, respectively. To our best knowledge, MSMM is the first classifier that Supports multiclass classification for Matrix-form EEG data. The proposed MSMM enables easier and more efficient implementation of robust multi-task BCIs, and therefore has potential to promote the wider use of BCI technology.
-
robust Support Matrix machine for single trial eeg classification
International Conference of the IEEE Engineering in Medicine and Biology Society, 2018Co-Authors: Qingqing Zheng, Fengyuan Zhu, Phengann HengAbstract:Electroencephalogram (EEG) signals are of complex structure and can be naturally represented as matrices. Classification is one of the most important steps for EEG signal processing. Newly developed classifiers can handle these Matrix-form data by adding low-rank constraint to leverage the correlation within each data. However, classification of EEG signals is still challenging, because EEG signals are always contaminated by measurement artifacts, outliers, and non-standard noise sources. As a result, existing Matrix classifiers may suffer from performance degradation, because they typically assume that the input EEG signals are clean. In this paper, to account for intra-sample outliers, we propose a novel classifier called a robust Support Matrix machine (RSMM), for single trial EEG data in Matrix form. Inspired by the fact that empirical EEG signals contain strong correlation information, we assume that each EEG Matrix can be decomposed into a latent low-rank clean Matrix plus a sparse noise Matrix. We simultaneously perform signal recovery and train the classifier based on the clean EEG matrices. We formulate our RSMM in a unified framework and present an effective solver based on the alternating direction method of multipliers. To evaluate the proposed method, we conduct extensive classification experiments on real binary EEG signals. The experimental results show that our method has outperformed the state-of-the-art Matrix classifiers. This paper may lead to the development of robust brain–computer interfaces (BCIs) with intuitive motor imagery and thus promote the broad use of the noninvasive BCIs technology.
Junsheng Cheng - One of the best experts on this subject based on the ideXlab platform.
-
symplectic weighted sparse Support Matrix machine for gear fault diagnosis
Measurement, 2021Co-Authors: Yu Yang, Haidong Shao, Xiang Zhong, Jian Cheng, Junsheng ChengAbstract:Abstract For gear fault diagnosis, it is often encountered that the input samples are naturally constructed as two-dimensional feature matrices with rich structure information. Support Matrix machine (SMM) is an effective classifier for these Matrix data, which fully leverages the Matrix structure information. However, it is indispensable for SMM to artificially extract fault features and select the useful features, which requires plenty of professional knowledge. Hence, a symplectic weighted sparse SMM (SWSSMM) model is proposed in this paper. Under the concept of symplectic geometry, SWSSMM automatically extracts the symplectic weighted coefficient Matrix (SWCM) as the fault feature representation. Meanwhile, the sparsity constraint and low-rank constraint are used in SWSSMM to eliminate the redundant fault features and capture the geometry structure information of SWCM, respectively. Besides, we derive an effective solver for SWSSMM with fast convergence. The experiment results demonstrate the superiority of SWSSMM for gear fault diagnosis.
-
symplectic incremental Matrix machine and its application in roller bearing condition monitoring
Applied Soft Computing, 2020Co-Authors: Yu Yang, Haiyang Pan, Ping Wang, Jian Wang, Junsheng ChengAbstract:Abstract For roller bearing condition monitoring, the collected signals have complex internal structure, which can be naturally represented as matrices. Support Matrix machine (SMM), as a new classifier with matrices as inputs, makes full use of the correlation between rows and columns of matrices and achieves ideal classification results. Unfortunately, SMM have ignored the issue of redundant features, which seriously affects the operational efficiency and recognition accuracy of algorithm. In this paper, we introduce symplectic geometry, l 1 -norm and incremental proximal descent (IPD) to SMM, and symplectic incremental Matrix machine (SIMM) is proposed. In SIMM, through symplectic geometry similarity transformation, the de-noising symplectic geometry coefficient Matrix is obtained, and the noise robustness of SMM method is therefore improved. Moreover, l 1 -norm is used to constrain the objective function, which can weaken the influence of redundant features, and thus greatly improving the recognition accuracy of SMM. Meanwhile, we use IPD to solve the objective function, which can obviously enhance the algorithm efficiency under the constant recognition rates. The experimental results of two kinds of roller bearings show that the proposed method has a good effectiveness in roller bearing condition monitoring, and the achieved recognition rate can reach 3%–25% much higher than those of the traditional recognition methods in 5-cross validation.
-
symplectic interactive Support Matrix machine and its application in roller bearing condition monitoring
Neurocomputing, 2020Co-Authors: Yu Yang, Haiyang Pan, Jinde Zheng, Junsheng ChengAbstract:Abstract Support Matrix machine (SMM) is an effective method to solve the problem of mechanical condition monitoring while the Matrix is taken as the input. It makes full use of the effective information between rows and columns of the Matrix to establish an ideal prediction model and achieve good condition monitoring results. However, Similar to Support vector machine (SVM), the core principle of the SMM is to distinguish the data effectively by two parallel hyperplanes. Unfortunately, two parallel hyperplanes may not be able to maximize the interval. Therefore, the concept of interactive Support Matrix machine (ISMM) is proposed, which constructs a pair of interactive hyperplanes to maximize the interval between two types of data. Interactive hyperplanes may be more able to distinguish between two types of data, so that each hyperplane is as close as to one of the two types and as far away as possible from the other. However, the input of the model often contain noise information, which seriously interferes with the classification results. Therefore, a symplectic interactive Support Matrix machine (SISMM) method is further proposed, which combines symplectic geometry similarity transformation (SGST) with ISMM. In SISMM, it can directly get the symplectic geometry coefficient Matrix without noise from the original signal, and intelligent classification recognition is realized. By analyzing and comparing the signal of roller bearings, the results show that the proposed method has better recognition performance and it is feasible for roller bearing condition monitoring.
-
a fault diagnosis approach for roller bearing based on symplectic geometry Matrix machine
Mechanism and Machine Theory, 2019Co-Authors: Yu Yang, Haiyang Pan, Jinde Zheng, Junsheng ChengAbstract:Abstract In many classification problems such as roller bearing fault diagnosis, it is often met that input samples are two-dimensional matrices constructed by vibration signals, and the rows or columns in the input matrices are strongly correlated. Support Matrix machine (SMM) is a new classifier with Matrix as input, which has a good diagnostic effect by using of Matrix structural information. Unfortunately, SMM algorithm is essentially binary, which need carry on the multiple binary classifications for multi-class classification problem. Meanwhile, SMM method has limitations in dealing with the complex input matrices, such as noise robustness and convergence problem. Therefore, a new classification method, called symplectic geometry Matrix machine (SGMM), is proposed in this paper. In SGMM, by using symplectic geometry similarity transformation, the proposed method not only protects the original structure of the signal, but also automatically extracts noiseless features to establish weight coefficient model, which can achieve multi-class tasks. Meanwhile, because of establishment of weight coefficient model, the convergence problem can be avoided. The roller bearing fault signals are used to demonstrate the validity of the SGMM method, and the analysis results indicate that the proposed method has a good effectiveness in roller bearing fault diagnosis.
