L1/2正则子在Lq (0 < q < 1)正则子中的代表性: 基于相位图的实验研究

徐宗本; 郭海亮; 王尧; 张海

doi:10.3724/SP.J.1004.2012.01225

L_1/2正则子在L_q (0 < q < 1)正则子中的代表性: 基于相位图的实验研究

doi: 10.3724/SP.J.1004.2012.01225

计量
- 文章访问数: 1884
- HTML全文浏览量: 65
- PDF下载量: 1611
- 被引次数: 0
出版历程
- 收稿日期: 2011-03-16
- 修回日期: 2011-06-22
- 刊出日期: 2012-07-20

Representative of L_1/2 Regularization among L_q (0 < q ≤ 1) Regularizations: an Experimental Study Based on Phase Diagram

摘要

摘要: 近期, 正则化方法吸引了越来越多的关注. 在L1正则子之后，Lq (0 q 1) 正则子被提出用于更好的求解稀疏性问题. 一个自然的问题是：在所有Lq (0 q 1) 正则子中, 哪一个q是最好的选择？通过采用相位图, 以及一组关于信号恢复与误差校正问题的实验, 我们表明: (i) 随着q减小, Lq正则子得到更稀疏的解; (ii) 当1/2L1/2正则子始终产生最好的稀疏解，且当0 q 1/2时，正则子的性能没有显著的区别. 因此, 我们认为L1/2正则子可被看作是一个Lq (0 q 1) 正则子的代表.
- Lq正则子 /
- 相位图 /
- 信号恢复 /
- 误差校正
Abstract: Recently, regularization methods have attracted in-creasing attention. Lq (0 q 1) regularizations were proposed after L1 regularization for better solution of sparsity problems. A natural question is which is the best choice among Lq regu-larizations with all q in (0, 1)? By taking phase diagram studies with a set of experiments implemented on signal recovery and error correction problems, we show the following: 1) As the value of q decreases, the Lq regularization generates sparser solution. 2) When 1/2 ≤ q 1, the L1/2 regularization always yields the best sparse solution and when 0 q ≤ 1/2, the performance of the regularizatons takes no significant difference. Accordingly, we conclude that the L1/2 regularization can be taken as a rep-resentative of Lq (0 q 1) regularizations.
- Lq regularization /
- phase diagram /
- signal recovery /
- error correction

HTML全文

参考文献(1)

[1]

Natarajan B K. Sparse approximate solutions to linear systems. SIAM Journal on Computing, 1995, 24(2): 227-234[2] Chen S, Donoho D L, Saunders M A. Atomic decomposition by basis pursuit. SIAM Journal on Scientific Computing, 1998, 20(1): 33-61[3] Candes E, Wakin M, Boyd S. Enhancing sparsity by reweighted L_1 minimization. Journal of Fourier Analysis and Applications, 2008, 14(5): 877-905[4] Chartrand R. Exact reconstructions of sparse signals via nonconvex minimization. IEEE Signal Processing Letters, 2007, 14(10): 707-710[5] Chartrand R, Staneva V. Restricted isometry properties and nonconvex compressive sensing. Inverse Problems, 2008, 24(3): 1-14[6] Donoho D L. Neighborly polytopes and the sparse solution of underdetermined linear equations [Online], available: http: // www-stat.stanford.edu/ ～ donoho/ Reports/ 2005/ NPaSSULE-01-28-05.pdf, November 8, 2011[7] Donoho D L. High-dimensional centrally symmetric polytopes with neighborliness proportional to dimension. Discrete and Computational Geometry, 2006, 35(4): 617-652[8] Donoho D L, Stodden V. Breakdown point of model selection when the number of variables exceeds the number of observations. In: Proceedings of the International Joint Conference on Neural Networks. Vancouver, USA: IEEE, 2006. 1916-1921[9] Xu Z B, Zhang H, Wang Y, Chang X Y, Yong L. L_1/2 regularization. Science in China Series F: Information Sciences, 2010, 53(6): 1159-1169[10] Chen X, Xu F M, Ye Y. Lower bound theory of nonzero entries in solutions of L_2-L_p minimization. SIAM Journal on Scientific Computing, 2010, 32(5): 2832-2852[11] Chartrand R, Yin W. Iteratively reweighted algorithms for compressive sensing. In: Proceedings of the IEEE International Conference on Acoustics, Speech and Signal Processing. Las Vegas, USA: IEEE, 2008. 3869-3872[12] Candes E, Tao T. Decoding by linear programming. IEEE Transactions on Information Theory, 2005, 51(12): 4203-4215[13] Chartrand R. Nonconvex compressed sensing and error correction. In: Proceedings of the IEEE International Conference on Acoustics, Speech and Signal Processing. Honolulu, USA: IEEE, 2007. 889-892[14] Candes E, Caltech J R. L_1-MAGIC: recovery of sparse signals via convex programming [Online], available: http: //users.ece.gatech.edu/～ justin/l1magic/downloads/l1magic. pdf, June 30, 2011[15] Donoho D L, Tanner J. Observed universality of phase transitions in high-dimensional geometry, with implications for modern data analysis and signal processing. Philosophical Transactions of the Royal Society A, 2009, 367(1906): 4273-4293

施引文献

资源附件(0)

访问统计