论文部分内容阅读
A fast feature ranking algorithm for classification in the presence of high dimensionality and small sample size is proposed.The basic idea is that the important features force the data points of the same class to maintain their intrinsic neighbor rela