Simple Classification Models for Coarse-Grained Problem

There exist datasets which cannot achieve high accuracy for all the classification models in the current application. In view of this problem, this paper proposed a high precision coarse-grained classification model. The main idea of coarse-grained model is to predict a set of labels that might contain the gold label. Experiments show that the model is effective in small data sets and can help people to make decisions. We summarized the model and discussed possible future improvements.