Hongzong Li

New method removes assignment bias in counterfactual prediction without adversarial training

Researchers have developed a new method for counterfactual prediction that avoids adversarial training by using information-regularized representations. This approach aims to remove treatment-covariate dependence by lea…