大规模核方法的随机假设空间方法-计算机科学与探索.PDFVIP

ISSN 1673-9418 CODEN JKYTA8 E-mail: fcst@vip. 163.com Journal of Frontiers of Computer Science and Technology 1673-9418/2018/ 12(05)-0785-09 Tel: +86-10 doi: 10.3778/j.issn. 1673-9418.1705027 大规模核方法的随机假设空间方法* 冯 昌，廖士中+ 天津大学 计算机科学与技术学院，天津 300350 Large-Scale Kernel Methods via Random Hypothesis Spaces FENG Chang, LIAO Shizhong+ School of Computer Science and Technology, Tianjin University, Tianjin 300350, China + Corresponding author: E-mail: szliao@ FENG Chang, LIAO Shizhong. Large-scale kernel methods via random hypothesis spaces. Journal of Fron- tiers of Computer Science and Technology, 2018, 12(5) ：785-793. Abstract: Large-scale kernel methods are main machine learning methods for current big data analysis and mining. Kernel methods search the reproducing kernel Hilbert spaces (RKHSs) for some hypotheses to handle original non- linear learning problems, which requires quadratic time complexity with respect to the size of training set and makes predicting with a kernelized procedure on the entire training set. Hence kernel methods are unpractical for large- scale datasets. To address these issues, this paper proposes an efficient approach to large-scale kernel methods with 索 random hypothesis spaces. Firstly, the circulant random feature mapping is used to construct random hypothesis 探 spaces, called circulant random hypothesis spaces (CRHSs), in loglinear time with respect to the dimensionality of input data. Then, in CRHSs, off-the-shelf linear algorithms are applied to train linear model, which has linear or sub-