Efficient Evaluation of Prediction Rules in Semi-Supervised Settings under Stratified Sampling
分层抽样下半监督预测规则的高效评估
Abstract In many contemporary applications, large amounts of unlabelled data are readily available while labelled examples are limited. There has been substantial interest in semi-supervised learning (SSL) which aims to leverage unlabelled data to improve estimation or prediction. However, current SSL literature focuses primarily on settings where labelled data are selected uniformly at random from the population of interest. Stratified sampling, while posing additional…