Target Aggregation Regression based on Random Forests

Abstract

In the era of big data, data volume is growing explosively, and the proportion of unlabeled data is increasing due to the high cost of label acquisition and privacy protection policy. Therefore, weakly labeled learning has drawn much attention to recognize patterns in the real-world data. Learning with target aggregation is an emerging technique in the last few years. However, to the best of our knowledge, the target aggregation regression problem has not been discussed. In this paper, we make use of random forest to build regression model for data with aggregated targets. The experiment shows the promising result of the proposed model.