Estimated error rates for different training models.
收藏数据链接：
官方服务：
资源简介：
Type I and II error breakdown for various training regimes. (Top) Classifiers trained with varying approaches to handling ambiguous cases can be effectively positioned along their ROC curve. (Bottom) When data is further broken into easy and hard cases, easy examples see moderately increased agreement and hard examples see more disagreement, resembling the human agreement histogram of Figure 4.
创建时间：
2015-12-02
