Pub Date : 2020-06-01DOI: 10.1109/SAM48682.2020.9104289
Lan Lan, Angela Marino, A. Aubry, A. Maio, G. Liao, Jingwei Xu
Adaptive target detection in the presence of Gaussian interference considering a frequency diverse array multipleinput multiple-output (FDA-MIMO) radar is investigated. At the design stage, adaptive detectors are devised according to the generalized likelihood ratio test (GLRT) criterion, where the target range is assumed unknown within the radar cell. Hence, the maximum likelihood (ML) estimate of the target range under the H1 hypothesis is approximated either assuming a discrete grid or resorting to a Newton-based optimization procedure. Numerical results are provided to illustrate the effectiveness of the devised detectors.
{"title":"Design of adaptive detectors for FDA-MIMO radar","authors":"Lan Lan, Angela Marino, A. Aubry, A. Maio, G. Liao, Jingwei Xu","doi":"10.1109/SAM48682.2020.9104289","DOIUrl":"https://doi.org/10.1109/SAM48682.2020.9104289","url":null,"abstract":"Adaptive target detection in the presence of Gaussian interference considering a frequency diverse array multipleinput multiple-output (FDA-MIMO) radar is investigated. At the design stage, adaptive detectors are devised according to the generalized likelihood ratio test (GLRT) criterion, where the target range is assumed unknown within the radar cell. Hence, the maximum likelihood (ML) estimate of the target range under the H1 hypothesis is approximated either assuming a discrete grid or resorting to a Newton-based optimization procedure. Numerical results are provided to illustrate the effectiveness of the devised detectors.","PeriodicalId":6753,"journal":{"name":"2020 IEEE 11th Sensor Array and Multichannel Signal Processing Workshop (SAM)","volume":"63 1","pages":"1-5"},"PeriodicalIF":0.0,"publicationDate":"2020-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"86545714","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-06-01DOI: 10.1109/SAM48682.2020.9104279
Yotam Gigi, Sella Nevo, G. Elidan, A. Hassidim, Yossi Matias, A. Wiesel
We consider multiple matrix regression tasks that share common weights in order to reduce sample complexity. For this purpose, we introduce the common mechanism regression model which assumes a shared right low-rank component across all tasks, but allows an individual per-task left low-rank component. We provide a closed form spectral algorithm for recovering the common component and derive a bound on its error as a function of the number of related tasks and the number of samples available for each of them. Both the algorithm and its analysis are natural extensions of known results in the context of phase retrieval and low rank reconstruction. We demonstrate the efficacy of our approach for the challenging task of remote river discharge estimation across multiple river sites, where data for each task is naturally scarce. In this scenario sharing a low-rank component between the tasks translates to a shared spectral reflection of the water, which is a true underlying physical model. We also show the benefit of the approach in the setting of image classification where the common component can be interpreted as the shared convolution filters.
{"title":"Spectral Algorithm for Shared Low-rank Matrix Regressions","authors":"Yotam Gigi, Sella Nevo, G. Elidan, A. Hassidim, Yossi Matias, A. Wiesel","doi":"10.1109/SAM48682.2020.9104279","DOIUrl":"https://doi.org/10.1109/SAM48682.2020.9104279","url":null,"abstract":"We consider multiple matrix regression tasks that share common weights in order to reduce sample complexity. For this purpose, we introduce the common mechanism regression model which assumes a shared right low-rank component across all tasks, but allows an individual per-task left low-rank component. We provide a closed form spectral algorithm for recovering the common component and derive a bound on its error as a function of the number of related tasks and the number of samples available for each of them. Both the algorithm and its analysis are natural extensions of known results in the context of phase retrieval and low rank reconstruction. We demonstrate the efficacy of our approach for the challenging task of remote river discharge estimation across multiple river sites, where data for each task is naturally scarce. In this scenario sharing a low-rank component between the tasks translates to a shared spectral reflection of the water, which is a true underlying physical model. We also show the benefit of the approach in the setting of image classification where the common component can be interpreted as the shared convolution filters.","PeriodicalId":6753,"journal":{"name":"2020 IEEE 11th Sensor Array and Multichannel Signal Processing Workshop (SAM)","volume":"89 1","pages":"1-5"},"PeriodicalIF":0.0,"publicationDate":"2020-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"85603090","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-06-01DOI: 10.1109/SAM48682.2020.9104263
Junkang Wei, Jianing Li, Z. Cao, C. Qin, Chunyi Song, Xu Zhiwei
Passive radar has shown many advantages as compared to active radar. However, it also faces more challenges owing to the weak signal for detection, especially when using GEO satellite signals as illuminator of opportunity (IO). High processing gain is then required. In this paper, a radar prototype using the standard digital television signals of Digital Video Broadcast-Satellite (DVB-S) as IO is developed, which applies cross-correlation over three broadcasting channels of Apstar-6 simultaneously and jointly. Experiments are conducted to test performance of the prototype.
{"title":"A Passive Radar Prototype Based on Multi-channel Joint Detection and Its Test Results","authors":"Junkang Wei, Jianing Li, Z. Cao, C. Qin, Chunyi Song, Xu Zhiwei","doi":"10.1109/SAM48682.2020.9104263","DOIUrl":"https://doi.org/10.1109/SAM48682.2020.9104263","url":null,"abstract":"Passive radar has shown many advantages as compared to active radar. However, it also faces more challenges owing to the weak signal for detection, especially when using GEO satellite signals as illuminator of opportunity (IO). High processing gain is then required. In this paper, a radar prototype using the standard digital television signals of Digital Video Broadcast-Satellite (DVB-S) as IO is developed, which applies cross-correlation over three broadcasting channels of Apstar-6 simultaneously and jointly. Experiments are conducted to test performance of the prototype.","PeriodicalId":6753,"journal":{"name":"2020 IEEE 11th Sensor Array and Multichannel Signal Processing Workshop (SAM)","volume":"16 1","pages":"1-5"},"PeriodicalIF":0.0,"publicationDate":"2020-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"87842468","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-06-01DOI: 10.1109/SAM48682.2020.9104304
Yan Liu, Xue Jiang
Channel estimation is the process of estimating channel parameters from the received samples, which are corrupted by noise. Most of the conventional methods are designed for noise-free or Gaussian noise environment. However, impulsive noise, which is also referred to as outliers, is common in practice and performance of the conventional algorithms degrades in the presence of outliers. In this paper, we propose a robust super-resolution channel estimation algorithm to deal with outliers by replacing ℓ2-norm constraints with ℓ1-norm constraints to enhance robustness to outliers and solving an improved convex program to obtain the channel parameters, the angles and time delays then are estimated jointly. Simulation results show that the proposed robust super-resolution channel estimation algorithm outperforms the traditional methods and show great robustness to the outliers.
{"title":"Robust Super-resolution Frequency Division Duplex (FDD) Channel Estimation","authors":"Yan Liu, Xue Jiang","doi":"10.1109/SAM48682.2020.9104304","DOIUrl":"https://doi.org/10.1109/SAM48682.2020.9104304","url":null,"abstract":"Channel estimation is the process of estimating channel parameters from the received samples, which are corrupted by noise. Most of the conventional methods are designed for noise-free or Gaussian noise environment. However, impulsive noise, which is also referred to as outliers, is common in practice and performance of the conventional algorithms degrades in the presence of outliers. In this paper, we propose a robust super-resolution channel estimation algorithm to deal with outliers by replacing ℓ2-norm constraints with ℓ1-norm constraints to enhance robustness to outliers and solving an improved convex program to obtain the channel parameters, the angles and time delays then are estimated jointly. Simulation results show that the proposed robust super-resolution channel estimation algorithm outperforms the traditional methods and show great robustness to the outliers.","PeriodicalId":6753,"journal":{"name":"2020 IEEE 11th Sensor Array and Multichannel Signal Processing Workshop (SAM)","volume":"29 1","pages":"1-5"},"PeriodicalIF":0.0,"publicationDate":"2020-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"78276422","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Conventional direction of arrival (DOA) estimation methods for known waveform sources have often assumed the noise being Gaussian distribution. However, the impulsive noise are frequently encountered in practice, which may severely degrade their estimation performance. To deal with this problem, we firstly construct a hybrid cost function to obtain the matrix related to unknown DOAs and complex amplitudes in the presence of impulsive noise. Then, by incorporating the majorization-minimization (MM) framework, the cost function is optimized iteratively. Finally, the DOAs and complex amplitudes are estimated via using the inherent relationship of the matrix calculated from the MM step. As demonstrated by simulation results, for strongly impulsive noise, the proposed method has a better estimation performance than many existing methods. Moreover, it can handle both weakly and strongly impulsive cases effectively.
{"title":"Robust DOA Estimation for Sources with Known Waveforms in Impulsive Noise Environments","authors":"Yang-yang Dong, Chun-xi Dong, Zhongshan Wu, Jingjing Cai, Hua Chen","doi":"10.1109/SAM48682.2020.9104251","DOIUrl":"https://doi.org/10.1109/SAM48682.2020.9104251","url":null,"abstract":"Conventional direction of arrival (DOA) estimation methods for known waveform sources have often assumed the noise being Gaussian distribution. However, the impulsive noise are frequently encountered in practice, which may severely degrade their estimation performance. To deal with this problem, we firstly construct a hybrid cost function to obtain the matrix related to unknown DOAs and complex amplitudes in the presence of impulsive noise. Then, by incorporating the majorization-minimization (MM) framework, the cost function is optimized iteratively. Finally, the DOAs and complex amplitudes are estimated via using the inherent relationship of the matrix calculated from the MM step. As demonstrated by simulation results, for strongly impulsive noise, the proposed method has a better estimation performance than many existing methods. Moreover, it can handle both weakly and strongly impulsive cases effectively.","PeriodicalId":6753,"journal":{"name":"2020 IEEE 11th Sensor Array and Multichannel Signal Processing Workshop (SAM)","volume":"54 1","pages":"1-5"},"PeriodicalIF":0.0,"publicationDate":"2020-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"76934640","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-06-01DOI: 10.1109/SAM48682.2020.9104336
R. Gui, Wen-qin Wang, H. So, Can Cui
As an emerging array processing technique, frequency diverse array (FDA) differs from conventional phased-array in that it employs a frequency increment across the array elements. The use of frequency increment provides FDA with an array factor with joint range-angle dependency, which finds wide applications in joint range-angle target localization and range-dependent interference/clutter suppression. In the open literature, an ideal point-like target in far field is generally assumed for FDA signal modelling. However, the reflectivity characterization for more realistic targets, which are not ideally point-like and even extended in range and azimuth angle, has not been reported for FDA radar. In this paper, we establish an echo signal model of FDA radar for a general target, and then analyze the statistics of the echo signal amplitude. More specifically, we reveal the amplitude decorrelation property between different FDA elements due to the use of frequency increment, which provides useful insight into frequency increment design. The target reflectivity characteristic analysis is validated by numerical results.
{"title":"Target Reflectivity Characterization for FDA Radar","authors":"R. Gui, Wen-qin Wang, H. So, Can Cui","doi":"10.1109/SAM48682.2020.9104336","DOIUrl":"https://doi.org/10.1109/SAM48682.2020.9104336","url":null,"abstract":"As an emerging array processing technique, frequency diverse array (FDA) differs from conventional phased-array in that it employs a frequency increment across the array elements. The use of frequency increment provides FDA with an array factor with joint range-angle dependency, which finds wide applications in joint range-angle target localization and range-dependent interference/clutter suppression. In the open literature, an ideal point-like target in far field is generally assumed for FDA signal modelling. However, the reflectivity characterization for more realistic targets, which are not ideally point-like and even extended in range and azimuth angle, has not been reported for FDA radar. In this paper, we establish an echo signal model of FDA radar for a general target, and then analyze the statistics of the echo signal amplitude. More specifically, we reveal the amplitude decorrelation property between different FDA elements due to the use of frequency increment, which provides useful insight into frequency increment design. The target reflectivity characteristic analysis is validated by numerical results.","PeriodicalId":6753,"journal":{"name":"2020 IEEE 11th Sensor Array and Multichannel Signal Processing Workshop (SAM)","volume":"9 1","pages":"1-5"},"PeriodicalIF":0.0,"publicationDate":"2020-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"90133631","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-06-01DOI: 10.1109/SAM48682.2020.9104345
Dima Taleb, M. Pesavento
In multicasting networks, a multi-antenna base station transmits the same information to a single group of users. In this work we consider general rank beamforming using orthogonal space-time block codes (OSTBC)s. The beamforming problem is non-convex and generally NP hard. The semidefinite relaxation technique is employed to solve the problem. In order to control the rank of the beamforming solution we propose to replace the power minimization by a regularized volume minimization which is known as a surrogate for the rank minimization. We propose an iterative two scale algorithm to find the appropriate value for the regularization parameter that yields the desired rank and to compute stationary points of the corresponding optimization problem. The high computational complexity of the proposed algorithm is improved significantly using a one scale algorithm, where the value of the regularization variable is reduced along with the decreasing rank. Simulation results demonstrate that our algorithms outperform the stateof-the-arts procedures in terms of the transmitted power and symbol error rate (SER). For a proper setting of the regularization variable, one scale algorithm outperforms the best compared methods in terms of computational complexity.
{"title":"Rank regularized beamforming in single group multicasting networks","authors":"Dima Taleb, M. Pesavento","doi":"10.1109/SAM48682.2020.9104345","DOIUrl":"https://doi.org/10.1109/SAM48682.2020.9104345","url":null,"abstract":"In multicasting networks, a multi-antenna base station transmits the same information to a single group of users. In this work we consider general rank beamforming using orthogonal space-time block codes (OSTBC)s. The beamforming problem is non-convex and generally NP hard. The semidefinite relaxation technique is employed to solve the problem. In order to control the rank of the beamforming solution we propose to replace the power minimization by a regularized volume minimization which is known as a surrogate for the rank minimization. We propose an iterative two scale algorithm to find the appropriate value for the regularization parameter that yields the desired rank and to compute stationary points of the corresponding optimization problem. The high computational complexity of the proposed algorithm is improved significantly using a one scale algorithm, where the value of the regularization variable is reduced along with the decreasing rank. Simulation results demonstrate that our algorithms outperform the stateof-the-arts procedures in terms of the transmitted power and symbol error rate (SER). For a proper setting of the regularization variable, one scale algorithm outperforms the best compared methods in terms of computational complexity.","PeriodicalId":6753,"journal":{"name":"2020 IEEE 11th Sensor Array and Multichannel Signal Processing Workshop (SAM)","volume":"28 1","pages":"1-5"},"PeriodicalIF":0.0,"publicationDate":"2020-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"90523610","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-06-01DOI: 10.1109/SAM48682.2020.9104296
Tianyao Long, Yong Jia, Li Jiang, Binge Yan, Tanzheng Yang
This paper mainly deals with the problem of direction-of-arrival (DOA) estimation for the ultra sparse nested (USN) MIMO array by operating on two co-prime frequencies. The USN MIMO array consists of a sparse uniform (SU) transmitting array and a more SU receiving array with nested relationship, which generates a SU sum coarray. In this case, the DOA estimation is aliasing because the difference coarray of the sum coarray (DCSC) of the USN MIMO array is also SU for the reference operation frequency. To remove the aliasing, an additional operation frequency with co-prime relationship is utilized to form an extra SU sum coarray where the spacing of two adjacent virtual sensors is co-prime with that of reference frequency(RF). As a result, two co-prime spacings of sum coarrays are combined into a co-prime sum coarray which provides a desired DCSC with a majority of contiguous virtual sensors. Finally, with respect to these contiguous virtual DCSC sensors, an augmented correlation matrix with contiguous correlation lags is obtained to calculate MUSIC spectrum. Simulation results demonstrate the resolvable ability for more targets than physical sensors and the performance comparison under both cases of proportional and non-propotional target spectra.
{"title":"DOA Estimation Based on Ultra Sparse Nested MIMO Array with Two Co-prime Frequencies","authors":"Tianyao Long, Yong Jia, Li Jiang, Binge Yan, Tanzheng Yang","doi":"10.1109/SAM48682.2020.9104296","DOIUrl":"https://doi.org/10.1109/SAM48682.2020.9104296","url":null,"abstract":"This paper mainly deals with the problem of direction-of-arrival (DOA) estimation for the ultra sparse nested (USN) MIMO array by operating on two co-prime frequencies. The USN MIMO array consists of a sparse uniform (SU) transmitting array and a more SU receiving array with nested relationship, which generates a SU sum coarray. In this case, the DOA estimation is aliasing because the difference coarray of the sum coarray (DCSC) of the USN MIMO array is also SU for the reference operation frequency. To remove the aliasing, an additional operation frequency with co-prime relationship is utilized to form an extra SU sum coarray where the spacing of two adjacent virtual sensors is co-prime with that of reference frequency(RF). As a result, two co-prime spacings of sum coarrays are combined into a co-prime sum coarray which provides a desired DCSC with a majority of contiguous virtual sensors. Finally, with respect to these contiguous virtual DCSC sensors, an augmented correlation matrix with contiguous correlation lags is obtained to calculate MUSIC spectrum. Simulation results demonstrate the resolvable ability for more targets than physical sensors and the performance comparison under both cases of proportional and non-propotional target spectra.","PeriodicalId":6753,"journal":{"name":"2020 IEEE 11th Sensor Array and Multichannel Signal Processing Workshop (SAM)","volume":"38 5 1","pages":"1-5"},"PeriodicalIF":0.0,"publicationDate":"2020-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"89477440","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-06-01DOI: 10.1109/SAM48682.2020.9104340
Lu Gan, Hongqing Liu
We consider the problem of recovering a band-limited signal from multi-channel modulo folded measurements. The study is motivated by the recent advances of self-reset analog-to-digital converters (SR-ADCs) for high dynamic range digitization. Most of existing works focus on single-channel SR-ADC, which requires high sampling rate and the reconstruction could be very unstable at low signal to noise ratio. To our best knowledge, this is the first work that studies multi-channel SR-ADC systems. In the noiseless case, we show that perfect reconstruction can be achieved using only 2 channels, each of which samples at Nyquist rate. For noisy measurements, we develop a lattice-based optimization for stable reconstruction. Simulation results are presented to demonstrate the effectiveness of the proposed algorithms.
{"title":"High Dynamic Range Sensing Using Multi-Channel Modulo Samplers","authors":"Lu Gan, Hongqing Liu","doi":"10.1109/SAM48682.2020.9104340","DOIUrl":"https://doi.org/10.1109/SAM48682.2020.9104340","url":null,"abstract":"We consider the problem of recovering a band-limited signal from multi-channel modulo folded measurements. The study is motivated by the recent advances of self-reset analog-to-digital converters (SR-ADCs) for high dynamic range digitization. Most of existing works focus on single-channel SR-ADC, which requires high sampling rate and the reconstruction could be very unstable at low signal to noise ratio. To our best knowledge, this is the first work that studies multi-channel SR-ADC systems. In the noiseless case, we show that perfect reconstruction can be achieved using only 2 channels, each of which samples at Nyquist rate. For noisy measurements, we develop a lattice-based optimization for stable reconstruction. Simulation results are presented to demonstrate the effectiveness of the proposed algorithms.","PeriodicalId":6753,"journal":{"name":"2020 IEEE 11th Sensor Array and Multichannel Signal Processing Workshop (SAM)","volume":"36 1","pages":"1-5"},"PeriodicalIF":0.0,"publicationDate":"2020-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"89112981","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Data dimensionality reduction via linear embedding is a typical approach to economizing the computational cost of machine learning systems. In the context of sparse subspace clustering (SSC), this paper proposes a two-step neighbor identification scheme using linear neighborhoodpreserving embedding. In the first step, a quadratically- constrained ℓ1 -minimization algorithm is solved for acquiring the side subspace neighborhood information, whereby a linear neighborhood-preserving embedding is designed accordingly. In the second step, a LASSO sparse regression algorithm is conducted for neighbor identification using the dimensionality- reduced data. The proposed embedding design explicitly takes into account the subspace neighborhood structure of the given data set. Computer simulations using real human face data show that the proposed embedding not only outperforms other existing dimensionality-reduction schemes but also improves the global data clustering accuracy when compared to the baseline solution without data compression.
{"title":"Sparse Subspace Clustering with Linear Subspace-Neighborhood-Preserving Data Embedding","authors":"Jwo-Yuh Wu, L. Huang, Wen-Hsuan Li, Hau-Hsiang Chan, Chun-Hung Liu, Rung-Hung Gau","doi":"10.1109/SAM48682.2020.9104396","DOIUrl":"https://doi.org/10.1109/SAM48682.2020.9104396","url":null,"abstract":"Data dimensionality reduction via linear embedding is a typical approach to economizing the computational cost of machine learning systems. In the context of sparse subspace clustering (SSC), this paper proposes a two-step neighbor identification scheme using linear neighborhoodpreserving embedding. In the first step, a quadratically- constrained ℓ1 -minimization algorithm is solved for acquiring the side subspace neighborhood information, whereby a linear neighborhood-preserving embedding is designed accordingly. In the second step, a LASSO sparse regression algorithm is conducted for neighbor identification using the dimensionality- reduced data. The proposed embedding design explicitly takes into account the subspace neighborhood structure of the given data set. Computer simulations using real human face data show that the proposed embedding not only outperforms other existing dimensionality-reduction schemes but also improves the global data clustering accuracy when compared to the baseline solution without data compression.","PeriodicalId":6753,"journal":{"name":"2020 IEEE 11th Sensor Array and Multichannel Signal Processing Workshop (SAM)","volume":"9 1","pages":"1-5"},"PeriodicalIF":0.0,"publicationDate":"2020-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"79477612","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：