正在加载页面…
Figure 3 reports that test error decreases quickly as the number P of random binning partitions grows, while training-plus-testing time grows slowly; the authors conclude that good performance can be obtained with a modest number of features. · CiteArk