WSDMOct, 2020
因果传递随机森林:结合记录数据和随机实验进行强鲁棒性预测
Causal Transfer Random Forest: Combining Logged Data and Randomized Experiments for Robust Prediction
Shuxi Zeng, Murat Ali Bayir, Joesph J.Pfeiffer III, Denis Charles, Emre Kiciman
TL;DR提出了一种 causal transfer random forest 模型,它将原有的训练数据与来自一个随机实验的少量数据组合,使其对 feature shift 具有鲁棒性,并在点击预测等任务中表现出优越性。