基于更贪心策略的快速正交核匹配追踪算法  

Fast Orthogonal Kernel Matching Pursuit Based on Greedier Strategy

作　　者：付丽华[1] 李宏伟[1] 张猛[2] 

机构地区：[1]中国地质大学(武汉)数学与物理学院,湖北武汉430074 [2]华中师范大学计算机学院,湖北武汉430079

出　　处：《电子学报》2013年第8期1580-1585,共6页Acta Electronica Sinica

基　　金：国家自然科学基金(No.61102103;No.61071188);中央高校基本科研业务费专项资金(No.CUG110407;No.CUG120110;No.CCNU10A01013)

摘　　要：为了得到信号的稀疏表达结果,正交核匹配追踪采用贪婪算法,在逐步回归建模过程中,每步只寻求当前最优原子,这使得计算效率大大降低.针对此局限性,提出了一种新的更加贪心的策略:在每次回归时,选择超过阈值的一个或者多个原子.为了更好的提高算法精度和稀疏度,再利用原子相似度对所挑选的原子做进一步筛选.实验结果和计算复杂度的分析说明:较传统的方法,新的基于更贪心策略的方法不仅能够提高计算效率,而且所得到的模型具有稀疏性好,泛化能力高等优点.Orthogonal kernel matching pursuit（OKMP） for constructing sparse kernel models has been recently introduced,in which a greedy scheme is utilized to select a single element per iteration.To improve the efficiency and performance of the greedy-scheme-based OKMP,a greedier algorithm is considered.The main contribution is the development of a new selection strategy that effectively selects several elements in each iteration.The efficiency is achieved by reducing the regressor steps,thus the computation time of the orthogonalization that each newly selected regressors to all the selected terms before is saved.A pruned algorithm is proposed based on the similarity of the atoms to improve the accuracy of the approximation.Numerical results and computational complexity analysis show that this new scheme is capable of producing a much sparser regression model with better generalization than the conventional approaches.

关 键 词：更贪心策略 正交 核匹配追踪 稀疏 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]