Generalizability through Explainability: Countering Overfitting with Counterfactual Examples