An Approach to One-Bit Compressed Sensing Based on Probably Approximately Correct Learning Theory

被引:0
作者
Ahsen, Mehmet Eren [1 ]
Vidyasagar, Mathukumalli [2 ,3 ]
机构
[1] Icahn Sch Med Mt Sinai, 1468 Madison Ave, New York, NY 10029 USA
[2] Univ Texas Dallas, Dept Syst Engn, Richardson, TX 75080 USA
[3] Indian Inst Technol Hyderabad, Dept Elect Engn, Kandi 502285, Telangana, India
关键词
SIGNAL RECOVERY; NUMBER;
D O I
暂无
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
In this paper, the problem of one-bit compressed sensing (OBCS) is formulated as a problem in probably approximately correct (PAC) learning. It is shown that the Vapnik-Chervonenkis (VC-) dimension of the set of half-spaces in R-n generated by k-sparse vectors is bounded below by k(left perpendicular lg(n/k) right perpendicular + 1) and above by left perpendicular 2k lg(en) right perpendicular. By coupling this estimate with well-established results in PAC learning theory, we show that a consistent algorithm can recover a k-sparse vector with O(k lg n) measurements, given only the signs of the measurement vector. This result holds for all probability measures on R-n. The theory is also applicable to the case of noisy labels, where the signs of the measurements are flipped with some unknown probability.
引用
收藏
页数:23
相关论文
共 39 条
[1]   One-bit compressed sensing with non-Gaussian measurements [J].
Ai, Albert ;
Lapanowski, Alex ;
Plan, Yaniv ;
Vershynin, Roman .
LINEAR ALGEBRA AND ITS APPLICATIONS, 2014, 441 :222-239
[2]  
Angluin D., 1988, Machine Learning, V2, P343, DOI 10.1023/A:1022873112823
[3]  
[Anonymous], P AS C SIGN SYST COM
[4]  
[Anonymous], 2015, Sparse modeling: Theory, algorithms and applications
[5]  
[Anonymous], ARXIV171007973VA1
[6]  
[Anonymous], P C INF SCI SYST
[7]  
[Anonymous], P INT S INF THEOR
[8]  
Anthony Martin, 1999, Neural Network Learning
[9]   LEARNABILITY AND THE VAPNIK-CHERVONENKIS DIMENSION [J].
BLUMER, A ;
EHRENFEUCHT, A ;
HAUSSLER, D ;
WARMUTH, MK .
JOURNAL OF THE ACM, 1989, 36 (04) :929-965
[10]   Decoding by linear programming [J].
Candes, EJ ;
Tao, T .
IEEE TRANSACTIONS ON INFORMATION THEORY, 2005, 51 (12) :4203-4215