Yi Xu
Research Subjects
Yi Xu has a total of 11 co-authors. Here are the top co-authors by number of shared publications:
- Guiling Sun
College of Electronic Information and Optical Engineering, Nankai University, Tianjin 300350, China
2 shared publications - Jingfei He
Hebei university of Technology
2 shared publications Bowen Zheng
College of Electronic Information and Optical Engineering, Nankai University, Tianjin, China, 300071
1 shared publicationGuiling Sun
College of Electronic Information and Optical Engineering, Nankai University, Tianjin 300071 China
1 shared publication- Lijun Zhang
School of Materials Science and Engineering, Jilin University, Changchun, China
1 shared publication - Tianbao Yang
Texas A&M University
1 shared publication
Latest Publications
- Open Access
Stochastic Optimization for Non-convex Inf-Projection Problems
In this paper, we study a family of non-convex and possibly non-smooth inf-projection minimization problems, where the target objective function is equal to minimization of a joint function over another variable. This problem include difference of convex (DC) functions and a family of bi-convex functions as special cases. We develop stochastic algorithms and establish their first-order convergence for finding a (nearly) stationary solution of the target non-convex function under different conditions of the component functions. To the best of our knowledge, this is the first work that comprehensively studies stochastic optimization of non-convex inf-projection minimization problems with provable convergence guarantee. Our algorithms enable efficient stochastic optimization of a family of non-decomposable DC functions and a family of bi-convex functions. To demonstrate the power of the proposed algorithms we consider an important application in variance-based regularization. Experiments verify the effectiveness of our inf-projection based formulation and the proposed stochastic algorithm in comparison with previous stochastic algorithms based on the min-max formulation for achieving the same effect.
Open AccessStochastic Optimization for Non-convex Inf-Projection Problems
In this paper, we study a family of non-convex and possibly non-smooth inf-projection minimization problems, where the target objective function is equal to minimization of a joint function over another variable. This problem include difference of convex (DC) functions and a family of bi-convex functions as special cases. We develop stochastic algorithms and establish their first-order convergence for finding a (nearly) stationary solution of the target non-convex function under different conditions of the component functions. To the best of our knowledge, this is the first work that comprehensively studies stochastic optimization of non-convex inf-projection minimization problems with provable convergence guarantee. Our algorithms enable efficient stochastic optimization of a family of non-decomposable DC functions and a family of bi-convex functions. To demonstrate the power of the proposed algorithms we consider an important application in variance-based regularization. Experiments verify the effectiveness of our inf-projection based formulation and the proposed stochastic algorithm in comparison with previous stochastic algorithms based on the min-max formulation for achieving the same effect.
- Open Access
- 23 October 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 7, 155242-155250
Compressive Sparse Data Gathering With Low-Rank and Total Variation in Wireless Sensor Networks
Wireless Sensor Networks (WSNs) have been deeply studied by many researchers and been widely used in many fields. Since a large amount of energy for WSNs is used for sensing and transmitting, researchers come up with many methods to reduce the number of sensed and transmitted data packets. Compressive Data Gathering (CDG) is a well-known method to gather WSNs data, but it does not realize sparse sensing as it needs to sense all data and compress them. The efficiency of Low-rank and TV regularizations for recovering WSNs data has been demonstrated, however, they are not combined to enable utilization of data correlation throughout the network. To recover the data accurately and to reduce the energy consumption in WSNs, we propose a Compressive Sparse Data Gathering (CSDG) scheme including a Compressive Sparse Sampling (CSS) method and a data recovery algorithm based on low-rank and Total Variation (TV) regularizations fully exploiting the sparsity and low-rank characteristics of WSNs data. The alternating direction method of multipliers and the steepest gradient descent method are used to solve the problem. Simulations show that the CSDG method outperforms the state-of-the-art methods in terms of the recovery accuracy. Moreover, with fairly low sparse sampling ratio and high compression ratio, CSDG method can still recover the original signal with little error. As the number of sensed data and transmitted data is reduced greatly with sparse sampling and compression, the energy consumption of WSNs is lessen and the lifetime is prolonged.
- 23 October 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 7, 155242-155250
Open AccessCompressive Sparse Data Gathering With Low-Rank and Total Variation in Wireless Sensor Networks
Wireless Sensor Networks (WSNs) have been deeply studied by many researchers and been widely used in many fields. Since a large amount of energy for WSNs is used for sensing and transmitting, researchers come up with many methods to reduce the number of sensed and transmitted data packets. Compressive Data Gathering (CDG) is a well-known method to gather WSNs data, but it does not realize sparse sensing as it needs to sense all data and compress them. The efficiency of Low-rank and TV regularizations for recovering WSNs data has been demonstrated, however, they are not combined to enable utilization of data correlation throughout the network. To recover the data accurately and to reduce the energy consumption in WSNs, we propose a Compressive Sparse Data Gathering (CSDG) scheme including a Compressive Sparse Sampling (CSS) method and a data recovery algorithm based on low-rank and Total Variation (TV) regularizations fully exploiting the sparsity and low-rank characteristics of WSNs data. The alternating direction method of multipliers and the steepest gradient descent method are used to solve the problem. Simulations show that the CSDG method outperforms the state-of-the-art methods in terms of the recovery accuracy. Moreover, with fairly low sparse sampling ratio and high compression ratio, CSDG method can still recover the original signal with little error. As the number of sensed data and transmitted data is reduced greatly with sparse sampling and compression, the energy consumption of WSNs is lessen and the lifetime is prolonged.
- 13 September 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Communications Letters
- Vol. 23 (12) , 2417-2421
Compressive Multi-Attribute Data Gathering Using Hankel Matrix in Wireless Sensor Networks
For heterogeneous WSNs with various types of sensors, compressive data gathering method requires more measurements due to the increased multiple attributes. In this letter, a compressive multi-attribute data gathering method using low-rank Hankel matrix is proposed to reduce the required measurements and improve the recovery accuracy in heterogeneous WSNs. Beyond utilizing just the spatiotemporal correlation of the raw sensed data with compressed sensing, the proposed method further enforces the low-rank block Hankel matrix to exploit the inherent correlation among multi-attribute data. Experimental results demonstrate that the proposed method can significantly improve the recovery accuracy of multi-attribute data compared with the existing solutions in WSNs.
- 13 September 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Communications Letters
- Vol. 23 (12) , 2417-2421
Compressive Multi-Attribute Data Gathering Using Hankel Matrix in Wireless Sensor Networks
For heterogeneous WSNs with various types of sensors, compressive data gathering method requires more measurements due to the increased multiple attributes. In this letter, a compressive multi-attribute data gathering method using low-rank Hankel matrix is proposed to reduce the required measurements and improve the recovery accuracy in heterogeneous WSNs. Beyond utilizing just the spatiotemporal correlation of the raw sensed data with compressed sensing, the proposed method further enforces the low-rank block Hankel matrix to exploit the inherent correlation among multi-attribute data. Experimental results demonstrate that the proposed method can significantly improve the recovery accuracy of multi-attribute data compared with the existing solutions in WSNs.
- Open Access
Low-Energy Data Collection in Wireless Sensor Networks Based on Matrix Completion
Sparse sensing schemes based on matrix completion for data collection have been proposed to reduce the power consumption of data-sensing and transmission in wireless sensor networks (WSNs). While extensive efforts have been made to improve the recovery accuracy from the sparse samples, it is usually at the cost of running time. Moreover, most data-collection methods are difficult to implement with low sampling ratio because of the communication limit. In this paper, we design a novel data-collection method including a Rotating Random Sparse Sampling method and a Fast Singular Value Thresholding algorithm. With the proposed method, nodes are in the sleep mode most of the time, and the sampling ratio varies over time slots during the sampling process. From the samples, a corresponding algorithm with Nesterov technique is given to recover the original data accurately and fast. With two real-world data sets in WSNs, simulations verify that our scheme outperforms other schemes in terms of energy consumption, reconstruction accuracy, and rate. Moreover, the proposed sampling method enhances the recovery algorithm and prolongs the lifetime of WSNs.
Open AccessLow-Energy Data Collection in Wireless Sensor Networks Based on Matrix Completion
Sparse sensing schemes based on matrix completion for data collection have been proposed to reduce the power consumption of data-sensing and transmission in wireless sensor networks (WSNs). While extensive efforts have been made to improve the recovery accuracy from the sparse samples, it is usually at the cost of running time. Moreover, most data-collection methods are difficult to implement with low sampling ratio because of the communication limit. In this paper, we design a novel data-collection method including a Rotating Random Sparse Sampling method and a Fast Singular Value Thresholding algorithm. With the proposed method, nodes are in the sleep mode most of the time, and the sampling ratio varies over time slots during the sampling process. From the samples, a corresponding algorithm with Nesterov technique is given to recover the original data accurately and fast. With two real-world data sets in WSNs, simulations verify that our scheme outperforms other schemes in terms of energy consumption, reconstruction accuracy, and rate. Moreover, the proposed sampling method enhances the recovery algorithm and prolongs the lifetime of WSNs.