Hossein Mohebbi

Counterfactual Residual Data Augmentation for Regression

Hossein Mohebbi, Oliver Schulte, Ke Li, Pascal Poupart

Data-driven modeling in real-world regression tasks often suffers from limited training samples, high collection costs, and noisy observations. Inspired by the impact of data augmentation in vision and language, we propose a novel Counterfactual Residual Data Augmentation (CRDA)…Read more