摘要
针对现有直觉模糊核匹配追踪算法采用贪婪算法搜索最优基函数而导致学习时间过长的局限性,基于弱贪婪策略,提出一种随机直觉模糊核匹配追踪算法.该算法不需要保证每次迭代过程都能搜索到当前最优基函数,仅需要在原搜索空间随机抽取一个较小的核字典子集进行搜索来获得近似最优基函数,从而有效减少一次迭代过程的搜索空间,大大降低了算法的训练时间.仿真结果表明,所提出方法在保持识别精度相当的情况下,有效缩短了一次匹配追踪时间,计算效率明显提高,且所得模型具有稀疏性好、泛化能力高等优点.
In order to overcome the long learning time caused by searching optimal basic function data based on the greedy strategy from a redundant basis function dictionary for the intuitionistic fuzzy kernel matching pursuit（IFKMP）, the random intuitionistic fuzzy kernel matching pursuit algorithm based on the weak greedy strategy is proposed. Rather than getting the present optimal basic function in each search, the approximate optimal basic function can be obtained by searching a random kernel dictionary subset of the original searching space, so that the searching space of matching pursuit can be reduced,and the training time can be decreased greatly. Simulation results show that, compared with the conventional approaches,the proposed algorithm can decrease training time and improve calculation efficiency obviously leaving the classification accuracy almost unchanged, while the model has better sparsity and generalization.
出处
《控制与决策》
EI
CSCD
北大核心
2016年第11期1960-1966,共7页
Control and Decision
基金
国家自然科学基金项目(61272011
61309022)
陕西省自然科学青年基金项目(2013JQ8031)
关键词
直觉模糊集
核匹配追踪
弱贪婪算法
目标识别
intuitionistic fuzzy set
kernel matching pursuit
week greedy algorithm
target recognition
