Enhancing Counterfactual Classification Performance via Self-Training

Abstract

Unlike traditional supervised learning, in many settings only partial feedback is available. We may only observe outcomes for the chosen actions, but not the counterfactual outcomes associated with other alternatives. Such settings encompass a wide variety of applications including pricing, online marketing and precision medicine. A key challenge is that observational data are influenced by historical policies deployed in the system, yielding a biased data distribution. We approach this task as a domain adaptation problem and propose a self-training algorithm which imputes outcomes with categorical values for finite unseen actions in the observational data to simulate a randomized trial through pseudolabelling, which we refer to as Counterfactual Self-Training (CST). CST iteratively imputes pseudolabels and retrains the model. In addition, we show input consistency loss can further improve CST performance which is shown in recent theoretical analysis of pseudolabelling. We demonstrate the effectiveness of the proposed algorithms on both synthetic and real datasets.

Semi-Counterfactual Risk Minimization Via Neural Networks

“…However, this approach fails to generalize well as shown by Beygelzimer and Langford (2009). Gao et al (2022) proposed another direct oriented method for off-line policy learning using the self-training approaches in semi-supervised learning.…”