运用Rademacher复杂度估计1-范数软间隔分类器的错分风险,所得到风险的界因与具体的数据有关,故分类器能针对不同的数据提供更确切的界.
The misclassification risk of 1 - norm soft margin classifier can be setimated by using Rademacher complexity. Beacause the risk bound is data - dependent, the classifier then yields more accurate bound when various samples are given.